Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Traceability and quality assessment of Norway spruce (Picea abies (L.) H.Karst.) logs: the TreeTrace_spruce database

Traceability and quality assessment of Norway spruce (Picea abies (L.) H.Karst.) logs: the... Key message The TreeTrace_spruce database contains images and measurements of 100 Norway spruce (Picea abies (L.) H.Karst.) logs from Northeastern France, each about 4.5 m long. The image database includes RGB images of large and small ends of the logs and hyperspectral and computed tomography (CT ) images of wood discs sampled at both log ends. The 100 logs were also fully X‑ray scanned with a CT device for roundwoods and their top surface was scanned with a terrestrial LiDAR device. The measurements performed on discs include wood local density, growth ring widths and pith location. This database is complementary to another one ( TreeTrace_Douglas) resulting from the same ANR project TreeTrace, but if the objectives are similar, the protocols and conditions of acquisition are not the same for these two databases. TreeTrace_spruce dataset is available at https:// doi. org/ 10. 57745/ WKLTJI and associated metadata are available at https:// metad ata‑ afs. nancy. inra. fr/ geone twork/ srv/ fre/ catal og. searc h#/ metad ata/ cffee 2f1‑ 18e1‑ 4b53‑ 9f5b‑ 6cc4c 66f1c b8. Keywords Image analysis, RGB images, Computed tomography, Picea abies, Wood density, Growth ring width, Terrestrial LIDAR 1 Background In this article, we present the database called TreeTrace_ spruce. This database mainly contains RGB, hyper - spectral and CT images of Norway spruce (Picea abies (L.) H.Karst.) log ends. Images of untreated log ends (i.e.  freshly sawn without any further preparation like Handling Editor: Véronique Lesage. sanding and polishing) were collected in this project. *Correspondence: Wood quality features such as pith position, growth ring Fleur Longuetaud widths and local wood density were measured on discs fleur.longuetaud@inrae.fr Université de Lorraine, AgroParisTech, INRAE, Silva, 54000 Nancy, France sampled at both log ends. CT and LiDAR scanning were Department of Computer Sciences, University of Salzburg, performed directly on the logs. 5020 Salzburg, Austria The data were collected in the framework of the Holztechnikum Kuchl, 5431 Kuchl, Austria Université de Lorraine, CNRS, LORIA, F‑54000 Nancy, France TreeTrace project (ANR-17-CE10-0016). The choice of Department of Forest Products Technology and Timber Construction, species in the TreeTrace project is related to the fact University of Applied Sciences Salzburg, 5412 Puch‑Urstein, Austria that the species Norway spruce and Douglas fir are of Department of Forest Utilisation, FVA Baden‑ Württemberg, 79100 Freiburg, Germany major importance in the three countries participating © The Author(s) 2023. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. Longuetaud et al. Annals of Forest Science (2023) 80:9 Page 2 of 12 in the project. In France, they represent the larg- • Wood traceability along the forest wood chain, est standing volumes and the highest annual growth, from the forest to the sawmill, by image processing as well as the largest volumes produced by sawmills of cross-sections only, potentially based on tech- (FCBA 2022). In Austria and Germany, too, spruce niques such as those used for fingerprint recogni - is the most important wood species (BFW 2022) and tion. Douglas fir is one of the promising climate change spe - • Assessment of the wood quality based on cross- cies in both countries (BFW 2008). section analysis mainly, applicable in the forest, on This article and the corresponding database are a log sorting platform or at the sawmill. The infor - complementary to the article already published on mation available on a cross-section (geometric cen- the Douglas fir part of the project (Longuetaud et  al. tre, pith location, juvenile wood area, number and 2022a) and the associated database TreeTrace_Doug- width of the annual growth rings) is complemen- las fir (https:// doi. org/ 10. 15454/ YUNEGL). The gen - tary to what could be obtained from the analysis eral objectives described below are similar but the of the external log shape by laser sensors installed protocols, data types and acquisition conditions dif- at the entrance of the sawing line (in practice, only fer between the two databases and it was important to tapering information is generally used). distinguish the two samplings. The main differences between the two databases are as follows: These application cases share many aspects. A com- mon dataset for experimental validation can be • TreeTrace_Douglas: The database was collected used, ground-truth data established by annotating in an industrial context with RGB images taken images can be shared and many software components on the log yard of a sawmill and then on the saw- implementing preprocessing (e.g.  pith detection, ing line with an industrial camera installed for the cross-section segmentation, contrast optimization) project. Moreover, as this was a strong originality as well as feature extraction techniques (e.g.  annual of this database, boards were collected and sent to ring detection) can be developed jointly and shared the LaBoMaP laboratory (Arts et Métiers Sciences subsequently. et Technologies, Cluny) to be analysed in terms of In the literature, very few references propose algo- mechanical quality with the final objective of link - rithms for image processing of untreated log ends. ing the quality information obtained on the logs Some algorithms exist for images of sanded and pol- and then on the boards; ished cross-sections in the context of dendrochro- • TreeTrace_spruce: The database was collected at nology but these algorithms are not made to process the felling site and then at the FVA (Forest Research images of rough log end sections. First application on Institute Baden-Württemberg, Freiburg, Germany), untreated log end images is to segment the wood cross- with acquisition of terrestrial LiDAR images of the section in the entire image. The method proposed by surface of the logs and CT images on the whole Schraml and Uhl (2014) to segment wood cross-sec- length of the 4.5 m logs that were scanned through tions of spruce is a similarity-based region growing the Microtec X-ray scanner at FVA. This point con - algorithm and requires pith location. Their main objec - stitutes a strong originality of this database with the tive was to trace log ends from the forest to the sawmill possibility of linking the external observations with (Schraml et  al. 2015). More recently, neural networks the internal structure of the logs. In addition, hyper- were used to perform this task (Wimmer et  al. 2021a). spectral images are also available for this species. A second application is to estimate the pith location. Norell and Borgefors (2008) have estimated the pith In the context of climate change and strong pressure on location in order to further detect tree rings (Norell resources, the optimisation of the forest and wood sec- 2011). Schraml and Uhl (2013) and Kurdthongmee tors is a challenge. In order to make the best use of wood et  al. (2018) proposed algorithms for pith detection on resources and ensure supplies to industries in terms of digital images of untreated wood cross-sections. For both quality and quantity, it is necessary to be able to Schraml and Uhl (2013), the objective was to retrieve describe the quality of the resource as early as possible cross-section characteristics in order to trace log ends. in the processing chain and to develop the traceability of Recently, Deep Neural Networks (DNNs) were trained this resource. Moreover, in the context of illegal logging, in order to address the problem of the pith detection in there is an increasing need to trace logs. cross-section images of logs (Kurdthongmee 2020). The data collected from the spruce sample of the The work already carried out with the TreeTrace_ TreeTrace project can be used for several purposes: spruce database is listed in Section 5. L onguetaud et al. Annals of Forest Science (2023) 80:9 Page 3 of 12 2 Methods 2.1 L og sampling in the field The sampling was performed in a Norway spruce even- aged stand located near Corcieux, France (48.1968 N; 6.8869 E). The trees were approximately 55 years old. No information was available about the silvicultural his- tory of the stand. A clear cut was performed a few days before our sampling on September 24 and 25, 2018. Since the logs were left in piles, it was not possible to know which logs belonged to the same tree (Fig.  1). The logs came from different heights in the trees. We selected 100 logs of 4.5-m length, rejecting logs with very large vis- ible defects. The logs have been labelled E001 to E100. The number of rings counted at the log ends (see Sec - tion 2.8) allowed to estimate the height in the stem of the logs (Ravoajanahary et  al. 2022). We concluded that the logs came from three heights in the tree: approximately Fig. 2 Diameter of the discs as a function of their height level in the 0m, 4.5m and 9m. Consequently the corresponding discs tree and their location in the sample logs (large end or small end). sampled at both log ends (Section  2.8) came from four Height levels 1, 2, 3 and 4 correspond to heights 0m, 4.5m, 9m and 13.5m, respectively heights in the tree: approximately 0m, 4.5m, 9m and 13.5m (Fig. 2). log_end_images/RGB/Corcieux_Huawei. A 2.2 I mage acquisition in the field in Corcieux (set #1) 25-cm-long calibration scale of 1 cm × 1 cm black and RGB images of large and small log ends were taken using white squares was included in each photo (Fig. 3). two devices: a Panasonic FZ45 Lumix camera (f/3.2 aper- For a subset of 100 images (one image of the large end ture, focal length 4.5 mm) and a Huawei P8 Lite 2017 of each of the 100 logs taken with the Lumix device), smartphone (f/2.2 aperture, focal length 3.8 mm). At the pixel widths in millimeters are given by the variable least five images of each log end were taken with each pixelWidth_mm. For each of these 100 images, a man- device: Two images taken with the same orientation, ual processing was done to locate the pith (pithX_pixel two images taken after tilting the camera by around 45°, and pithY_pixel) and to delineate the outline of the then a last image after stapling the label on the cross- wood cross-section under-bark. The segmented images section (Fig.  3). In total, 1009 images (4320 × 3240 pix- are available in TreeTrace_spruce/log_end_ els) taken with the Lumix camera and 1013 images images/RGB/Corcieux_Lumix/segmented/ (3968 × 2976 pixels) taken with the Huawei smart- large_ends. For each image, the under-bark area of phone were obtained at Corcieux (Table  1). The images the wood cross-section in pixels was computed (areaUB_ (raw and with labels) are available on the data reposi- pixel). All the RGB image-level data are in the file tory in TreeTrace_spruce/log_end_images/ TreeTrace_spruce/tables/rgb_images.txt. RGB/Corcieux_Lumix and TreeTrace_spruce/ Fig. 1 Logs immediately after harvesting (on the left). The 100 selected logs ready for photographing and labelling (on the right) Longuetaud et al. Annals of Forest Science (2023) 80:9 Page 4 of 12 Fig. 3 RGB images of cross‑section E024B_le taken with Lumix device (first row) and Huawei device (second row) at Corcieux Table 1 Number of RGB images taken at each site and with each camera at both ends of the 100 sampled logs Site Set Camera Large end Small end Corcieux 1 Lumix camera Raw 403 406 Labels 100 100 Huawei smartphone Raw 411 402 Labels 100 100 Freiburg 2 Huawei smartphone Raw 270 290 Labels 100 100 Munich and Kuchl 3 Canon camera Raw 399 376 Sanded 644 658 2.3 Naming conventions 2.4 Wood disc sampling and image acquisition in Freiburg Labels were stapled at both ends with the log number fol- (set #2) lowed by a letter B or H to indicate the bottom (B) or top The logs were then delivered to the Forest Research (H) side of the log. But since the log taper was not always Institute of Baden-Württemberg (FVA) at Freiburg, Ger- very pronounced, several errors occurred: Logs E019, many, on September 26, 2018, by a transporter. On Sep- E027, E032, E057, E062, E071, E072, E075, E087, E089, tember 27 and 28, discs of about 5 cm thick were cut at E090, E094 were thus mislabelled with H on bottom side both ends of each log (200 discs in total). The discs were and B on top side. These wrong names are the ones that then kept fresh in plastic film. The measurements that appear on the RGB images taken after stapling the labels have been made on discs are presented in Section 2.8. (fifth column in Fig. 3). After refreshing the log ends by cutting the discs, a new To avoid confusion, all the logs have been renamed set of RGB images of both log cross-sections was taken in this database with suffixes _le and _se to indicate the with the Huawei smartphone. Several images were taken actual large end (bottom side) and small end (top side) of of each log end: One image with the label before cutting the log, respectively. For instance, large end of log E024 the disc, then one to three images of the fresh section after was renamed E024B_le and large end of log E019 (misla- cutting the disc by changing the orientation of the camera belled with H) was renamed E019H_le. (Fig. 4). In total, 760 images (3968 × 2976 pixels) taken with The image files were named with respect to site, cam - the Huawei smartphone were obtained at Freiburg (Table 1) era device, log name, log end and image number. For and are available on the data repository in TreeTrace_ instance,  corcieux_lumix_E024B_le_3.jpg refers to the spruce/log_end_images/RGB/Freiburg_Hua- third image of large end of log E024 taken with the Lumix wei. A 25-cm-long calibration scale of 1 cm × 1 cm black device at Corcieux site, and corcieux_lumix_E024B_le_ and white squares was included in each photo. label.jpg is the image of the same cross-section after sta- As it was done for the images taken at Corcieux (Sec- pling the label. tion  2.2), for a subset of 100 images (one image of the L onguetaud et al. Annals of Forest Science (2023) 80:9 Page 5 of 12 Fig. 4 RGB images of cross‑section E024B_le taken with Huawei device at FVA Baden‑ Württemberg (Freiburg). The first image (left) was taken before cutting the disc, the other three after cutting the disc large end of each of the 100 logs), the pixel widths in that were scanned together. Four scans were performed millimeters are given by the variable pixelWidth_mm. for each set (Fig. 5). Each scan took 15 min with a resolu- For each of these 100 images, a manual processing was tion of 20,000 points/turn such as the distance between done to locate the pith (pithX_pixel and pithY_pixel) two cloud points was 3 mm at 10 m of distance. The and to delineate the outline of the wood cross-section scene was restricted to the area around the logs. No noise under-bark. The segmented images are available in filtering was applied to the cloud points. TreeTrace_spruce/log_end_images/RGB/ Several large diameter spheres were used for helping to Freiburg_Huawei/segmented/large_ends. merge the obtained clouds of points and small diameter For each image, the under-bark area of the wood cross- spheres were placed on each log for helping the matching section in pixels was computed (areaUB_pixel). All between T-LiDAR scans and CT scans (Figs. 5 and 6). the corresponding RGB image-level data are in the file Figure  7 shows the preview of a T-LiDAR cloud of TreeTrace_spruce/tables/rgb_images.txt. points. It should be noted that the bark of the logs was The image files were named with respect to site, relatively damaged by the harvester and transporta- camera device, log name, log end and image number. tion. Point clouds registration was performed using Faro For instance, freiburg_huawei_E024B_le_3.jpg refers to Scene 5.4 software taking position 4 as reference. In the the third image of large end of log E024 taken with the end, five merged point clouds, one for each set of 20 logs, Huawei device in Freiburg. are available on the data repository in TreeTrace_ spruce/log_images/tlidar. The cloud files are 2.5 Terrestrial LiDAR scanning of logs in text format with XYZ coordinates (in meters, Z cor- After cutting the discs, the logs were scanned using a responding to the altitude) and a reflectance value (given T-LiDAR device (Faro Focus 3D X130). The logs were as an RGB triplet). The log_order_and_position. placed on the ground over two supports by sets of 20 logs txt file gives the position of the logs in each cloud and Fig. 5 Locations of the device for T‑LiDAR scanning Longuetaud et al. Annals of Forest Science (2023) 80:9 Page 6 of 12 Fig. 6 A set of logs ready for T‑LiDAR scanning and T ‑LiDAR device in place at location 3 Fig. 7 T‑LiDAR view of a set of 20 logs scanned from location 3 the disc identifiers corresponding to the log ends ori - ID. The reconstruction files are 8-bit greyscale multipage ented toward position 4 in Fig. 5. tiff-files, with each page representing one log slice. The resolution of each page is 768 by 768 pixels. A second 2.6 X ‑ray scanning of logs image dataset TreeTrace_spruce/log_images/ CT scanning was performed immediately after T-LiDAR xray/segmented contains a semantic segmentation scanning at FVA Baden-Württemberg (Fig.  8) with the of the features of each log, which were detected in the CT.Log (MiCROTEC s.r.l.) scanner as described in Stän- CT scans. The images are 8-bit greyscale multipage tiff- gle et al. (2015). files, too, with individual grey levels for each of the fol - The logs were scanned at a generator voltage of 180 kV lowing five features: Air (area outside log): 255, bark: 150; and a generator current of 14 mA. The scanning direction sapwood: 200; heartwood: 250; pith: 10; region of sound follows the log from the larger (butt) end to the smaller knot:100; region of dead knot: 50. The feature detec - (top) end. The CT images were reconstructed to an axial tion was based on the CT.Pro software package (Version resolution of 4.8 mm or 5 mm and a resolution of 1.107 Dez-2015). mm × 1.107 mm in the planes perpendicular to the axial direction. The nominal length of each log extractable 2.7 Hyperspectral and RGB imaging of the discs from the CT scan is shorter than the physical log length before and after sanding in Munich (set #3) as 11 slices on the butt end and 13 slices from the top The discs were then transported to Stemmer Imaging end were cut due to the layout of the scanner with metal (https:// www. stemm er- imagi ng. com/) in Munich, Ger- holders of high material density which interferes with the many. During transport and intermediate storage, the log density as such. The image files of the reconstruction discs where packed individually in plastic bags to avoid TreeTrace_spruce/log_images/xray/tif are surface cracks due to wood shrinkage and discoloration named in consecutive order according to the original log due to oxidation processes (see Fig. 9b). L onguetaud et al. Annals of Forest Science (2023) 80:9 Page 7 of 12 Fig. 8 Installation of a log for CT scanning At Stemmer Imaging, the fresh cut side of each disc resolution of 640 × 640 pixel was chosen. Each wood was scanned with two different multispectral line scan - disc was scanned with both cameras. For this purpose, ners and RGB images were taken. Note, that all images each disc was pushed through the system by hand and captured from the wood discs are mirrored versions of the speed was synchronized with a trigger. The hyper- the images captured from the log ends. spectral data, i.e. the translation from line scanning The hyperspectral scanning setup is shown in data to a hyperspectral cube, was performed by the Fig.  9a—halogen light was used for lighting. The first acquisition software Perception Studio. scanner was a so-called Specim FX10, which scans After multispectral scanning with the first camera, the the spectra between a wavelength from 445 to 983nm same side was captured using a Canon 70D, 35mm lens. with a bandwidth of approx. 3nm. The second scanner At least four images with different rotations were taken of was a Specim FX17 which provides scans between 990 each disc. and 1665nm also with a bandwidth of approx. 3nm. After hyperspectral scanning and RGB imaging the Therefore, the Specim FX10 uses mainly the visible rough discs were transported to Kuchl, Austria, where light (VIS) spectra and the Specim FX17 uses mainly each disc was sanded with a mesh width of 120 to reduce parts of the NIR spectra. For the scanning setup, a the influence of manual chainsaw cuts at scanning. This Fig. 9 The sensor system shows the line scanner Specim FX10 which was mounted on a metal frame and the slice is moved perpendicular to the scanned line manually. The stored samples were closed airtight to avoid surface changes. Each hyperspectral cube was cropped to reduce the massive amount of data Longuetaud et al. Annals of Forest Science (2023) 80:9 Page 8 of 12 mesh width was chosen as it gave good results for the wet Then, the discs were scanned in air-dried state with samples. In Kuchl RGB images were taken using the same the medical CT scanner described in Freyburger et  al. setup as in Munich—at least six images with different (2009). 202 CT images are available since two discs rotations were taken. were scanned in two pieces. CT images, raw (DICOM) In total, 2077 images (4320 × 3240 pixels) taken with and calibrated in density (following the method the Canon camera at Munich and Kuchl were collected described in Freyburger et  al. (2009)), are provided (Table 1). The images are available on the data repository on the data repository in TreeTrace_spruce/ in TreeTrace_spruce/log_end_images/RGB/ log_end_images/xray. Munich_Canon. CT images were then processed as described in the The hyperspectral data (captured with the FX10 or Image processing section of Longuetaud et  al. (2016) FX17 camera) from each disc were stored in a special for- with the exception that only air-dried state was con- mat (HSD – hyperspectral data) utilized by Perception sidered. This method makes it possible to output air- Studio. A HSD file contains all spectral bands of a disc in dried-specific gravity in tangential bands of fixed mean kind of a hyperspectral cube. It was necessary to convert width and divided in several azimuthal sectors for the the data to the ENVI format for which open source image analysis of radial variations for example. Here, the mean processing libraries are available. width of the tangential bands was set to 1mm and the The ENVI files for the 200 discs are provided in the repos - bands were not divided azimuthally. A radial profile of itory under the following path: TreeTrace_spruce/ air-dried density consisting of the average distance to log_end_images/hyperspectral. the pith (meanRadius_mm) and density (airDryDen- sity_kgpm3) of each tangential band is thus obtained for each disc (TreeTrace_spruce/tables/discs_ 2.7.1 Notes on the hyperspectral data density_profiles.txt). Figure  10 provides exemplary images for one wood disc The radial profile of ring widths was coupled with the (#E001B). In the first and second row RGB images taken radial profile of density to assess the mean density of each from the raw and sanded disc surface are shown. In the annual growth ring (airDryDensity_kgpm3). The variable subsequent rows, for each hyperspectral camera, images was added to the disc-ring-level data file. The method is for selected bands are illustrated for the same disc, raw described in Ravoajanahary et al. (2022). and sanded. These disc-ring-level data were used to compute the It can be recognized that the width of the wood disc for data table at the disc level (TreeTrace_spruce/ the FX10 and FX17 images differs. We assume that this is tables/discs_measurements.txt). In this table, caused by the acquisition software and an inaccurate trig- the average density of a disc (airDryDensity_kgpm3) is ger synchronization. Furthermore, a set of cubes got lost the average of all pixels corresponding to wood (exclud- during file storage and are thus missing in the repository. ing bark) in the calibrated CT image. The number of Table  2 shows an overview of the number of captured these pixels was used to compute the under-bark area cubes for each hyperspectral camera. For the FX10 and of the discs (areaUBScanner_cm2). The quadratic mean the raw discs all cubes are available. of the four radii on which ring widths were measured (meanRadiusUBOptical_mm) is also reported. The 2.8 M easurements on wood discs height in the stem of each disc (estimatedHeight_m) was Finally, the 200 discs were transported to INRAE estimated based on the number of rings as explained in Nancy, France, in March 2019, for further analyses. Section 2.1. Ring widths were measured on four orthogonal radii and are provided in the disc-radius-ring-level data file (TreeTrace_spruce/tables/discs_radii_ 3 Access to the data and metadata description rings.txt). From these values, the average profile The images and the database (Longuetaud et  al. 2022b) of annual ring widths of a disc was obtained by sub- are available at Recherche Data Gouv repository: tracting the successive quadratic means of the four https:// doi. org/ 10. 57745/ WKLTJI. Associated metadata radii (i.e. root mean square of external radii of a given access is at https:// metad ata- afs. nancy. inra. fr/ geone ring minus root mean square of internal radii of the twork/ srv/ fre/ catal og. searc h#/ metad ata/ cffee 2f1- 18e1- same ring). The corresponding disc-ring-level data file 4b53- 9f5b- 6cc4c 66f1c b8. (TreeTrace_spruce/tables/discs_rings. The arborescence of the TreeTrace_spruce directory txt) provides cambial age (cambialAge), growth year was detailed in the previous section and is summarized (growthYear), distance to pith (externalRadius_mm) in Fig. 11. and ring width (ringWidth_mm). L onguetaud et al. Annals of Forest Science (2023) 80:9 Page 9 of 12 Fig. 10 Hyperspectral images of cross‑section E001B_le taken with FX10 camera (first row) and FX17 camera (second row) and RGB images taken with Canon device (third row) at Munich The main numeric data are in the TreeTrace_ is accompanied with a variable table (with the same spruce/tables directory. There are five tables, all name followed by _variable) describing each vari- in the form of an array with variables in columns and a able. The data tables are the following with the different first line indicating the variable names. Each data table levels of measurements: Longuetaud et al. Annals of Forest Science (2023) 80:9 Page 10 of 12 Table 2 Number of hyperspectral cubes taken with each • The discs_measurements table contains 200 camera in rough and sanded states at both ends of the 100 lines, one for each disc or log end, and 7 columns. sampled logs • The rgb_images table contains 1164 lines, one for each image on which any measurement has been Hyperspectral Surface Large end Small end camera made, and 8 columns. • The discs_rings table contains 8494 lines, one FX10 Raw 100 100 for each annual growth ring of each disc, and 7 col- FX10 Sanded 88 12 umns. The variables were computed from ring widths FX17 Raw 99 96 measured on orthogonal radii and density profiles FX17 Sanded 78 9 issued from CT images. • The discs_radii_rings table contains 8494 × 4 radii = 33976 lines, one for each ring measurement on each radius, and 6 columns. • The discs_density_profiles table contains 25,230 lines, one for each 1-mm-wide tangential Fig. 11 Organisation of the data files L onguetaud et al. Annals of Forest Science (2023) 80:9 Page 11 of 12 band on which air-dry density was measured on the the site and logs sampled since the trees and logs were CT images, and 5 columns. already cut when we arrived for sampling. We give below some examples of past uses of the database: An automatic reading algorithm of the black and white 4 Technical validation calibration scale was designed by Delconte (2019). All measurements were carefully checked by graphical Decelle and Jalilian (2020) proposed a neural network analyses with R software. The ring measurements were algorithm for automatic segmentation of the wood controlled by using a binocular magnifying glass when in cross-section in the raw images taken in the forest or doubt and by checking for consistency between ring pro- after transportation to Freiburg. files from the four radii measured on each disc. The X-ray The superiority of convolutional neural networks for CT scanner used to measure wood density is controlled the identification of log cross-sections over methods and calibrated by the manufacturer three times a year. adapted from fingerprint or iris recognition techniques A specific calibration procedure (Freyburger et  al. 2009) was demonstrated in the comparative study by Wim- was used to convert Hounsfield Units to wood density. mer et al. (2021a). Convolutional neural networks were also used by 5 Reuse potential and limits Wimmer et  al. (2021b) to identify a log from images All data tables are in plain text format with tabulations as taken by an RGB camera and a CT scanner. separator. They can be imported easily with R or with any Wimmer et  al. (2022) shows that the use of hyper- spreadsheet software. Sometimes information is missing spectral imaging and the combination of several spectra for some variables, which is translated in “NA” value in provides significantly better results than RGB images the tables. alone regarding the visibility of growth rings, which are The RGB images are in standard JPEG or PNG formats. the main characteristic used for traceability. A scale included in each image allows to perform spa- Ravoajanahary et  al. (2022) proposed a method for tial calibration. For a subset of RGB images (one image estimating tree ring density by coupling X-ray CT scan- per log taken at the large end and per site and device), ning and high-resolution optical measurements of tree the black and white scale was manually measured in the ring width. This method was applied to analyse the images and the image resolution is provided in a Table. relationship between ring density and ring width and The CT images of the discs (in the TreeTrace_ cambial age at several height levels in the trees. spruce/log_end_images/xray directory) and logs (in the TreeTrace_spruce/log_images/xray Acknowledgements Thanks to Forêt et Bois de l’Est for helping us by providing the 100 spruce logs directory) can be read with ImageJ software (https:// imagej. and by organizing log transportation to Freiburg. Thanks also to Martin Huber nih. gov/ ij/). from Forest Research Institute of Baden-Württemberg (FVA) and all the team for The registered T-LiDAR clouds (in the TreeTrace_ helping us with the organization of log measurement and imaging at FVA. spruce/log_images/tlidar directory) can be Authors’ contributions visualised with the 3D point cloud processing software FL initiated the funding request; managed the project for the French part; par‑ CloudCompare (https:// www. danie lgm. net/ cc/). ticipated in the data collection, the data analysis and writing the manuscript; and supervised PhD and Master students. RS initiated the funding request, did The hyperspectral images (in the TreeTrace_ his PhD thesis on the basis of these data and participated in the data collec‑ spruce/log_end_images/hyperspectral direc- tion, the data analysis and writing the manuscript. FM initiated the funding tory) require Matlab to be read. request and participated in the data collection, the data analysis and writing the manuscript. TR participated in the data collection, the data analysis and The image database TreeTrace_spruce, providing both writing the manuscript. RD did his PhD thesis on the basis of these data and raw images and ground truth, can be used to develop participated in the data collection, the data analysis and writing the manu‑ and to validate image analysis algorithms for detecting script. TC participated in the data collection, the data analysis and writing the manuscript. PN supervised the PhD and Master students and participated in various features (e.g. pith location, cross-sectional area, the data analysis. IDR supervised the PhD and Master students and partici‑ annual ring widths) on untreated (i.e.  freshly cut) cross- pated in the data analysis. KE supervised the PhD and Master students and sections of logs. participated in the data analysis. AP supervised the PhD and Master students and participated in the data analysis. FB participated in the data collection and The main objectives for the further development of writing the manuscript. AU initiated the funding request, managed the project algorithms are (1) to trace the logs from the forest to the for the Austrian part, participated in the data collection and supervised the sawmill and/or (2) to estimate the wood quality directly PhD and Master students. All authors read and approved the final manuscript. in the forest, on the log yard or on the sawmill line before Funding sawing. Both were objectives of the TreeTrace project. SilvaTech facility is supported by the French National Research Agency The difficulty in such studies performed in an indus - through the Laboratory of Excellence ARBRE (ANR‑11‑LABX ‑0002‑01). This research was made possible thanks to the financial support of the French trial context is that very few information is available on Longuetaud et al. Annals of Forest Science (2023) 80:9 Page 12 of 12 National Research Agency (ANR) and the Austrian Science Fund (FWF) in the Norell K (2011) Automatic counting of annual rings on Pinus sylvestris end framework of the TreeTrace project, ANR‑17‑ CE10‑0016. faces in sawmill industry. Comput Electron Agric 75(2):231–237. https:// doi. org/ 10. 1016/j. compag. 2010. 11. 005 Availability of data and materials Norell K, Borgefors G (2008) Estimation of pith position in untreated log ends The images and the database are available at Research Data Gouv repository: in sawmill environments. Comput Electron Agric 63(2):155–167. https:// https:// doi. org/ 10. 57745/ WKLTJI.doi. org/ 10. 1016/j. compag. 2008. 02. 006 Ravoajanahary T, Mothe F, Longuetaud F (2022) A method for estimating tree Code availability ring density by coupling CT scanning and ring width measurements: Not applicable. Application to the analysis of the ring width ‑ ring density relationship in Picea abies trees. Trees ‑ Structure and Function. https:// doi. org/ 10. 1007/ s00468‑ 022‑ 02373‑2 Declarations Schraml R, Hofbauer H, Petutschnigg A, Uhl A (2015) Tree log identification based on digital cross‑section images of log ends using fingerprint and Ethics approval and consent to participate iris recognition methods. In: Computer Analysis of Images and Patterns. Not applicable. CAIP 2015. Lecture Notes in Computer Science, vol 9256. Springer, pp 752–765. https:// doi. org/ 10. 1007/ 978‑3‑ 319‑ 23192‑1_ 63 Consent for publication Schraml R, Uhl A (2013) Pith estimation on rough log end images using local Not applicable. fourier spectrum analysis. In: Proceedings of the 14th Conference on Computer Graphics and Imaging (CGIM’13), Innsbruck, AUT. https:// doi. Competing interests org/ 10. 2316/P. 2013. 797‑ 012 The authors declare that they have no competing interests. Schraml R, Uhl A (2014) Similarity based crosssec ‑ tion segmentation in rough log end images. In: Artificial Intelligence Applications and Innovations. AIAI 2014. IFIP Advances in Information and Communication Technology, vol Received: 20 October 2022 Accepted: 25 January 2023 436. Springer, pp 614–623. https:// doi. org/ 10. 1007/ 9783‑662‑ 44654‑ 6_‑ 61 Stängle SM, Brüchert F, Heikkila A, Usenius T, Usenius A, Sauter UH (2015) Potentially increased sawmill yield from hardwoods using x‑ray com‑ puted tomography for knot detection. Ann For Sci 72(1):57–65. https:// doi. org/ 10. 1007/ s13595‑ 014‑ 0385‑1 References Wimmer G, Schraml R, Hofbauer H, Petutschnigg A, Uhl A (2021a) Two‑stage BFW (2008) Bundesforschungszentrum für wald. (2008) bfw‑praxisinformation cnn‑based wood log recognition. In: Computational Science and Its nr. 16 ‑ 2008. https:// bfw. ac. at/ 030/ pdf/ 1818_ pi16. pdf . Accessed 11 Jan Applications. ICCSA 2021. Lecture Notes in Computer Science, vol 12955. Springer, pp 115–125. https:// doi. org/ 10. 1007/ 978‑3‑ 030‑ 87007‑2_9 BFW (2022) Bundesforschungszentrum für wald. (2022) Österreichische Wimmer G, Schraml R, Hofbauer H, Petutschnigg A, Uhl A (2022) An analysis waldinventur. periode 2016/21. https:// www. waldi nvent ur. at. Accessed of the use of hyperspectral data for roundwood tracking. In: The 22nd 11 Jan 2023. International Conference on Computational Science and Its Applications Decelle R, Jalilian E (2020) Neural networks for cross‑section segmentation in (ICCSA 2022). Malaga, https:// doi. org/ 10. 1007/ 978‑3‑ 031‑ 10545‑6_ 21 raw images of log ends. In: 2020 IEEE 4th International Conference on Wimmer G, Schraml R, Lamminger L, Petutschnigg A, Uhl A (2021b) Cross‑ Image Processing, Applications and Systems (IPAS). IEEE, pp 131–137. modality wood log tracing. In: 2021 IEEE International Symposium on https:// doi. org/ 10. 1109/ IPAS5 0080. 2020. 93349 60 Multimedia (ISM). IEEE, pp 191–195. https:// doi. org/ 10. 1109/ ISM52 913. Delconte F (2019) Estimation de la qualité de grumes de bois. Master’s thesis, M2 2021. 00038 informatique, Parcours type Apprentissage, Vision, Robotique (AVR), Univer‑ sité de Lorraine, Nancy, France. https:// hal. univlor‑ ra ine. fr/ hal03603‑ 108 Publisher’s Note FCBA (2022) Memento 2022. p 48. https:// www. fcba. fr/ wp‑ conte nt/ uploa ds/ Springer Nature remains neutral with regard to jurisdictional claims in pub‑ 2023/ 01/ Memen to‑ 2022 WEB‑. pdf . Accessed 11 Jan 2023. lished maps and institutional affiliations. Freyburger C, Longuetaud F, Mothe F, Constant T, Leban JM (2009) Measur‑ ing wood density by means of x‑ray computer tomography. Ann For Sci 66(8):804. https:// doi. org/ 10. 1051/ forest/ 20090 71 Kurdthongmee W (2020) A comparative study of the effectiveness of using popular dnn object detection algorithms for pith detection in cross‑ sectional images of parawood. Heliyon 6(2). https:// doi. org/ 10. 1016/j. heliy on. 2020. e03480 Kurdthongmee W, Suwannarat K, Panyuen P, Sae‑Ma N (2018) A fast algorithm to approximate the pith location of rubberwood timber from a normal camera image. In: 15th International Joint Conference on Computer Sci‑ ence and Software Engineering (JCSSE). IEEE, pp 1–6. https:// doi. org/ 10. 1109/ JCSSE. 2018. 84573 75 Longuetaud F, Mothe F, Fournier M, Dlouha J, Santenoise P, Deleuze C Re Read ady y to to submit y submit your our re researc search h ? Choose BMC and benefit fr ? Choose BMC and benefit from om: : (2016) Within‑stem maps of wood density and water content for characterization of species: a case study on three hardwood and two fast, convenient online submission softwood species. Ann For Sci 73(3):601–614. https:// doi. org/ 10. 1007/ s13595‑ 016‑ 0555‑4 thorough peer review by experienced researchers in your field Longuetaud F, Pot G, Mothe F, Barthelemy A, Decelle R, Delconte F, Ge X, rapid publication on acceptance Guillaume G, Mancini T, Ravoajanahary T, Butaud JC, Collet R, Debled‑ support for research data, including large and complex data types Rennesson I, Marcon B, Ngo P, Roux B, Viguier J (2022a) Traceability and quality assessment of Douglas fir (Pseudotsuga menziesii (Mirb.) Franco) • gold Open Access which fosters wider collaboration and increased citations logs: the TreeTrace_Douglas database. Ann For Sci 79:46. https:// doi. org/ maximum visibility for your research: over 100M website views per year 10. 1186/ s13595‑ 022‑ 01163‑7 Longuetaud F, Schraml R, Mothe F, Ravoajanahary T, Decelle R, Constant T, At BMC, research is always in progress. Ngo P, Debled‑Rennesson I, Entacher K, Petutschnigg A, Brüchert F, Uhl A (2022b) “TreeTrace_spruce”. [dataset]. Recherche Data Gouv. Repository, Learn more biomedcentral.com/submissions V1. https:// doi. org/ 10. 57745/ WKLTJI http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Annals of Forest Science Springer Journals

Traceability and quality assessment of Norway spruce (Picea abies (L.) H.Karst.) logs: the TreeTrace_spruce database

Download PDF
12 pages

Loading next page...
 
/lp/springer-journals/traceability-and-quality-assessment-of-norway-spruce-picea-abies-l-h-qQaAxw8jUR
Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2023
ISSN
1286-4560
eISSN
1297-966X
DOI
10.1186/s13595-023-01178-8
Publisher site
See Article on Publisher Site

Abstract

Key message The TreeTrace_spruce database contains images and measurements of 100 Norway spruce (Picea abies (L.) H.Karst.) logs from Northeastern France, each about 4.5 m long. The image database includes RGB images of large and small ends of the logs and hyperspectral and computed tomography (CT ) images of wood discs sampled at both log ends. The 100 logs were also fully X‑ray scanned with a CT device for roundwoods and their top surface was scanned with a terrestrial LiDAR device. The measurements performed on discs include wood local density, growth ring widths and pith location. This database is complementary to another one ( TreeTrace_Douglas) resulting from the same ANR project TreeTrace, but if the objectives are similar, the protocols and conditions of acquisition are not the same for these two databases. TreeTrace_spruce dataset is available at https:// doi. org/ 10. 57745/ WKLTJI and associated metadata are available at https:// metad ata‑ afs. nancy. inra. fr/ geone twork/ srv/ fre/ catal og. searc h#/ metad ata/ cffee 2f1‑ 18e1‑ 4b53‑ 9f5b‑ 6cc4c 66f1c b8. Keywords Image analysis, RGB images, Computed tomography, Picea abies, Wood density, Growth ring width, Terrestrial LIDAR 1 Background In this article, we present the database called TreeTrace_ spruce. This database mainly contains RGB, hyper - spectral and CT images of Norway spruce (Picea abies (L.) H.Karst.) log ends. Images of untreated log ends (i.e.  freshly sawn without any further preparation like Handling Editor: Véronique Lesage. sanding and polishing) were collected in this project. *Correspondence: Wood quality features such as pith position, growth ring Fleur Longuetaud widths and local wood density were measured on discs fleur.longuetaud@inrae.fr Université de Lorraine, AgroParisTech, INRAE, Silva, 54000 Nancy, France sampled at both log ends. CT and LiDAR scanning were Department of Computer Sciences, University of Salzburg, performed directly on the logs. 5020 Salzburg, Austria The data were collected in the framework of the Holztechnikum Kuchl, 5431 Kuchl, Austria Université de Lorraine, CNRS, LORIA, F‑54000 Nancy, France TreeTrace project (ANR-17-CE10-0016). The choice of Department of Forest Products Technology and Timber Construction, species in the TreeTrace project is related to the fact University of Applied Sciences Salzburg, 5412 Puch‑Urstein, Austria that the species Norway spruce and Douglas fir are of Department of Forest Utilisation, FVA Baden‑ Württemberg, 79100 Freiburg, Germany major importance in the three countries participating © The Author(s) 2023. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. Longuetaud et al. Annals of Forest Science (2023) 80:9 Page 2 of 12 in the project. In France, they represent the larg- • Wood traceability along the forest wood chain, est standing volumes and the highest annual growth, from the forest to the sawmill, by image processing as well as the largest volumes produced by sawmills of cross-sections only, potentially based on tech- (FCBA 2022). In Austria and Germany, too, spruce niques such as those used for fingerprint recogni - is the most important wood species (BFW 2022) and tion. Douglas fir is one of the promising climate change spe - • Assessment of the wood quality based on cross- cies in both countries (BFW 2008). section analysis mainly, applicable in the forest, on This article and the corresponding database are a log sorting platform or at the sawmill. The infor - complementary to the article already published on mation available on a cross-section (geometric cen- the Douglas fir part of the project (Longuetaud et  al. tre, pith location, juvenile wood area, number and 2022a) and the associated database TreeTrace_Doug- width of the annual growth rings) is complemen- las fir (https:// doi. org/ 10. 15454/ YUNEGL). The gen - tary to what could be obtained from the analysis eral objectives described below are similar but the of the external log shape by laser sensors installed protocols, data types and acquisition conditions dif- at the entrance of the sawing line (in practice, only fer between the two databases and it was important to tapering information is generally used). distinguish the two samplings. The main differences between the two databases are as follows: These application cases share many aspects. A com- mon dataset for experimental validation can be • TreeTrace_Douglas: The database was collected used, ground-truth data established by annotating in an industrial context with RGB images taken images can be shared and many software components on the log yard of a sawmill and then on the saw- implementing preprocessing (e.g.  pith detection, ing line with an industrial camera installed for the cross-section segmentation, contrast optimization) project. Moreover, as this was a strong originality as well as feature extraction techniques (e.g.  annual of this database, boards were collected and sent to ring detection) can be developed jointly and shared the LaBoMaP laboratory (Arts et Métiers Sciences subsequently. et Technologies, Cluny) to be analysed in terms of In the literature, very few references propose algo- mechanical quality with the final objective of link - rithms for image processing of untreated log ends. ing the quality information obtained on the logs Some algorithms exist for images of sanded and pol- and then on the boards; ished cross-sections in the context of dendrochro- • TreeTrace_spruce: The database was collected at nology but these algorithms are not made to process the felling site and then at the FVA (Forest Research images of rough log end sections. First application on Institute Baden-Württemberg, Freiburg, Germany), untreated log end images is to segment the wood cross- with acquisition of terrestrial LiDAR images of the section in the entire image. The method proposed by surface of the logs and CT images on the whole Schraml and Uhl (2014) to segment wood cross-sec- length of the 4.5 m logs that were scanned through tions of spruce is a similarity-based region growing the Microtec X-ray scanner at FVA. This point con - algorithm and requires pith location. Their main objec - stitutes a strong originality of this database with the tive was to trace log ends from the forest to the sawmill possibility of linking the external observations with (Schraml et  al. 2015). More recently, neural networks the internal structure of the logs. In addition, hyper- were used to perform this task (Wimmer et  al. 2021a). spectral images are also available for this species. A second application is to estimate the pith location. Norell and Borgefors (2008) have estimated the pith In the context of climate change and strong pressure on location in order to further detect tree rings (Norell resources, the optimisation of the forest and wood sec- 2011). Schraml and Uhl (2013) and Kurdthongmee tors is a challenge. In order to make the best use of wood et  al. (2018) proposed algorithms for pith detection on resources and ensure supplies to industries in terms of digital images of untreated wood cross-sections. For both quality and quantity, it is necessary to be able to Schraml and Uhl (2013), the objective was to retrieve describe the quality of the resource as early as possible cross-section characteristics in order to trace log ends. in the processing chain and to develop the traceability of Recently, Deep Neural Networks (DNNs) were trained this resource. Moreover, in the context of illegal logging, in order to address the problem of the pith detection in there is an increasing need to trace logs. cross-section images of logs (Kurdthongmee 2020). The data collected from the spruce sample of the The work already carried out with the TreeTrace_ TreeTrace project can be used for several purposes: spruce database is listed in Section 5. L onguetaud et al. Annals of Forest Science (2023) 80:9 Page 3 of 12 2 Methods 2.1 L og sampling in the field The sampling was performed in a Norway spruce even- aged stand located near Corcieux, France (48.1968 N; 6.8869 E). The trees were approximately 55 years old. No information was available about the silvicultural his- tory of the stand. A clear cut was performed a few days before our sampling on September 24 and 25, 2018. Since the logs were left in piles, it was not possible to know which logs belonged to the same tree (Fig.  1). The logs came from different heights in the trees. We selected 100 logs of 4.5-m length, rejecting logs with very large vis- ible defects. The logs have been labelled E001 to E100. The number of rings counted at the log ends (see Sec - tion 2.8) allowed to estimate the height in the stem of the logs (Ravoajanahary et  al. 2022). We concluded that the logs came from three heights in the tree: approximately Fig. 2 Diameter of the discs as a function of their height level in the 0m, 4.5m and 9m. Consequently the corresponding discs tree and their location in the sample logs (large end or small end). sampled at both log ends (Section  2.8) came from four Height levels 1, 2, 3 and 4 correspond to heights 0m, 4.5m, 9m and 13.5m, respectively heights in the tree: approximately 0m, 4.5m, 9m and 13.5m (Fig. 2). log_end_images/RGB/Corcieux_Huawei. A 2.2 I mage acquisition in the field in Corcieux (set #1) 25-cm-long calibration scale of 1 cm × 1 cm black and RGB images of large and small log ends were taken using white squares was included in each photo (Fig. 3). two devices: a Panasonic FZ45 Lumix camera (f/3.2 aper- For a subset of 100 images (one image of the large end ture, focal length 4.5 mm) and a Huawei P8 Lite 2017 of each of the 100 logs taken with the Lumix device), smartphone (f/2.2 aperture, focal length 3.8 mm). At the pixel widths in millimeters are given by the variable least five images of each log end were taken with each pixelWidth_mm. For each of these 100 images, a man- device: Two images taken with the same orientation, ual processing was done to locate the pith (pithX_pixel two images taken after tilting the camera by around 45°, and pithY_pixel) and to delineate the outline of the then a last image after stapling the label on the cross- wood cross-section under-bark. The segmented images section (Fig.  3). In total, 1009 images (4320 × 3240 pix- are available in TreeTrace_spruce/log_end_ els) taken with the Lumix camera and 1013 images images/RGB/Corcieux_Lumix/segmented/ (3968 × 2976 pixels) taken with the Huawei smart- large_ends. For each image, the under-bark area of phone were obtained at Corcieux (Table  1). The images the wood cross-section in pixels was computed (areaUB_ (raw and with labels) are available on the data reposi- pixel). All the RGB image-level data are in the file tory in TreeTrace_spruce/log_end_images/ TreeTrace_spruce/tables/rgb_images.txt. RGB/Corcieux_Lumix and TreeTrace_spruce/ Fig. 1 Logs immediately after harvesting (on the left). The 100 selected logs ready for photographing and labelling (on the right) Longuetaud et al. Annals of Forest Science (2023) 80:9 Page 4 of 12 Fig. 3 RGB images of cross‑section E024B_le taken with Lumix device (first row) and Huawei device (second row) at Corcieux Table 1 Number of RGB images taken at each site and with each camera at both ends of the 100 sampled logs Site Set Camera Large end Small end Corcieux 1 Lumix camera Raw 403 406 Labels 100 100 Huawei smartphone Raw 411 402 Labels 100 100 Freiburg 2 Huawei smartphone Raw 270 290 Labels 100 100 Munich and Kuchl 3 Canon camera Raw 399 376 Sanded 644 658 2.3 Naming conventions 2.4 Wood disc sampling and image acquisition in Freiburg Labels were stapled at both ends with the log number fol- (set #2) lowed by a letter B or H to indicate the bottom (B) or top The logs were then delivered to the Forest Research (H) side of the log. But since the log taper was not always Institute of Baden-Württemberg (FVA) at Freiburg, Ger- very pronounced, several errors occurred: Logs E019, many, on September 26, 2018, by a transporter. On Sep- E027, E032, E057, E062, E071, E072, E075, E087, E089, tember 27 and 28, discs of about 5 cm thick were cut at E090, E094 were thus mislabelled with H on bottom side both ends of each log (200 discs in total). The discs were and B on top side. These wrong names are the ones that then kept fresh in plastic film. The measurements that appear on the RGB images taken after stapling the labels have been made on discs are presented in Section 2.8. (fifth column in Fig. 3). After refreshing the log ends by cutting the discs, a new To avoid confusion, all the logs have been renamed set of RGB images of both log cross-sections was taken in this database with suffixes _le and _se to indicate the with the Huawei smartphone. Several images were taken actual large end (bottom side) and small end (top side) of of each log end: One image with the label before cutting the log, respectively. For instance, large end of log E024 the disc, then one to three images of the fresh section after was renamed E024B_le and large end of log E019 (misla- cutting the disc by changing the orientation of the camera belled with H) was renamed E019H_le. (Fig. 4). In total, 760 images (3968 × 2976 pixels) taken with The image files were named with respect to site, cam - the Huawei smartphone were obtained at Freiburg (Table 1) era device, log name, log end and image number. For and are available on the data repository in TreeTrace_ instance,  corcieux_lumix_E024B_le_3.jpg refers to the spruce/log_end_images/RGB/Freiburg_Hua- third image of large end of log E024 taken with the Lumix wei. A 25-cm-long calibration scale of 1 cm × 1 cm black device at Corcieux site, and corcieux_lumix_E024B_le_ and white squares was included in each photo. label.jpg is the image of the same cross-section after sta- As it was done for the images taken at Corcieux (Sec- pling the label. tion  2.2), for a subset of 100 images (one image of the L onguetaud et al. Annals of Forest Science (2023) 80:9 Page 5 of 12 Fig. 4 RGB images of cross‑section E024B_le taken with Huawei device at FVA Baden‑ Württemberg (Freiburg). The first image (left) was taken before cutting the disc, the other three after cutting the disc large end of each of the 100 logs), the pixel widths in that were scanned together. Four scans were performed millimeters are given by the variable pixelWidth_mm. for each set (Fig. 5). Each scan took 15 min with a resolu- For each of these 100 images, a manual processing was tion of 20,000 points/turn such as the distance between done to locate the pith (pithX_pixel and pithY_pixel) two cloud points was 3 mm at 10 m of distance. The and to delineate the outline of the wood cross-section scene was restricted to the area around the logs. No noise under-bark. The segmented images are available in filtering was applied to the cloud points. TreeTrace_spruce/log_end_images/RGB/ Several large diameter spheres were used for helping to Freiburg_Huawei/segmented/large_ends. merge the obtained clouds of points and small diameter For each image, the under-bark area of the wood cross- spheres were placed on each log for helping the matching section in pixels was computed (areaUB_pixel). All between T-LiDAR scans and CT scans (Figs. 5 and 6). the corresponding RGB image-level data are in the file Figure  7 shows the preview of a T-LiDAR cloud of TreeTrace_spruce/tables/rgb_images.txt. points. It should be noted that the bark of the logs was The image files were named with respect to site, relatively damaged by the harvester and transporta- camera device, log name, log end and image number. tion. Point clouds registration was performed using Faro For instance, freiburg_huawei_E024B_le_3.jpg refers to Scene 5.4 software taking position 4 as reference. In the the third image of large end of log E024 taken with the end, five merged point clouds, one for each set of 20 logs, Huawei device in Freiburg. are available on the data repository in TreeTrace_ spruce/log_images/tlidar. The cloud files are 2.5 Terrestrial LiDAR scanning of logs in text format with XYZ coordinates (in meters, Z cor- After cutting the discs, the logs were scanned using a responding to the altitude) and a reflectance value (given T-LiDAR device (Faro Focus 3D X130). The logs were as an RGB triplet). The log_order_and_position. placed on the ground over two supports by sets of 20 logs txt file gives the position of the logs in each cloud and Fig. 5 Locations of the device for T‑LiDAR scanning Longuetaud et al. Annals of Forest Science (2023) 80:9 Page 6 of 12 Fig. 6 A set of logs ready for T‑LiDAR scanning and T ‑LiDAR device in place at location 3 Fig. 7 T‑LiDAR view of a set of 20 logs scanned from location 3 the disc identifiers corresponding to the log ends ori - ID. The reconstruction files are 8-bit greyscale multipage ented toward position 4 in Fig. 5. tiff-files, with each page representing one log slice. The resolution of each page is 768 by 768 pixels. A second 2.6 X ‑ray scanning of logs image dataset TreeTrace_spruce/log_images/ CT scanning was performed immediately after T-LiDAR xray/segmented contains a semantic segmentation scanning at FVA Baden-Württemberg (Fig.  8) with the of the features of each log, which were detected in the CT.Log (MiCROTEC s.r.l.) scanner as described in Stän- CT scans. The images are 8-bit greyscale multipage tiff- gle et al. (2015). files, too, with individual grey levels for each of the fol - The logs were scanned at a generator voltage of 180 kV lowing five features: Air (area outside log): 255, bark: 150; and a generator current of 14 mA. The scanning direction sapwood: 200; heartwood: 250; pith: 10; region of sound follows the log from the larger (butt) end to the smaller knot:100; region of dead knot: 50. The feature detec - (top) end. The CT images were reconstructed to an axial tion was based on the CT.Pro software package (Version resolution of 4.8 mm or 5 mm and a resolution of 1.107 Dez-2015). mm × 1.107 mm in the planes perpendicular to the axial direction. The nominal length of each log extractable 2.7 Hyperspectral and RGB imaging of the discs from the CT scan is shorter than the physical log length before and after sanding in Munich (set #3) as 11 slices on the butt end and 13 slices from the top The discs were then transported to Stemmer Imaging end were cut due to the layout of the scanner with metal (https:// www. stemm er- imagi ng. com/) in Munich, Ger- holders of high material density which interferes with the many. During transport and intermediate storage, the log density as such. The image files of the reconstruction discs where packed individually in plastic bags to avoid TreeTrace_spruce/log_images/xray/tif are surface cracks due to wood shrinkage and discoloration named in consecutive order according to the original log due to oxidation processes (see Fig. 9b). L onguetaud et al. Annals of Forest Science (2023) 80:9 Page 7 of 12 Fig. 8 Installation of a log for CT scanning At Stemmer Imaging, the fresh cut side of each disc resolution of 640 × 640 pixel was chosen. Each wood was scanned with two different multispectral line scan - disc was scanned with both cameras. For this purpose, ners and RGB images were taken. Note, that all images each disc was pushed through the system by hand and captured from the wood discs are mirrored versions of the speed was synchronized with a trigger. The hyper- the images captured from the log ends. spectral data, i.e. the translation from line scanning The hyperspectral scanning setup is shown in data to a hyperspectral cube, was performed by the Fig.  9a—halogen light was used for lighting. The first acquisition software Perception Studio. scanner was a so-called Specim FX10, which scans After multispectral scanning with the first camera, the the spectra between a wavelength from 445 to 983nm same side was captured using a Canon 70D, 35mm lens. with a bandwidth of approx. 3nm. The second scanner At least four images with different rotations were taken of was a Specim FX17 which provides scans between 990 each disc. and 1665nm also with a bandwidth of approx. 3nm. After hyperspectral scanning and RGB imaging the Therefore, the Specim FX10 uses mainly the visible rough discs were transported to Kuchl, Austria, where light (VIS) spectra and the Specim FX17 uses mainly each disc was sanded with a mesh width of 120 to reduce parts of the NIR spectra. For the scanning setup, a the influence of manual chainsaw cuts at scanning. This Fig. 9 The sensor system shows the line scanner Specim FX10 which was mounted on a metal frame and the slice is moved perpendicular to the scanned line manually. The stored samples were closed airtight to avoid surface changes. Each hyperspectral cube was cropped to reduce the massive amount of data Longuetaud et al. Annals of Forest Science (2023) 80:9 Page 8 of 12 mesh width was chosen as it gave good results for the wet Then, the discs were scanned in air-dried state with samples. In Kuchl RGB images were taken using the same the medical CT scanner described in Freyburger et  al. setup as in Munich—at least six images with different (2009). 202 CT images are available since two discs rotations were taken. were scanned in two pieces. CT images, raw (DICOM) In total, 2077 images (4320 × 3240 pixels) taken with and calibrated in density (following the method the Canon camera at Munich and Kuchl were collected described in Freyburger et  al. (2009)), are provided (Table 1). The images are available on the data repository on the data repository in TreeTrace_spruce/ in TreeTrace_spruce/log_end_images/RGB/ log_end_images/xray. Munich_Canon. CT images were then processed as described in the The hyperspectral data (captured with the FX10 or Image processing section of Longuetaud et  al. (2016) FX17 camera) from each disc were stored in a special for- with the exception that only air-dried state was con- mat (HSD – hyperspectral data) utilized by Perception sidered. This method makes it possible to output air- Studio. A HSD file contains all spectral bands of a disc in dried-specific gravity in tangential bands of fixed mean kind of a hyperspectral cube. It was necessary to convert width and divided in several azimuthal sectors for the the data to the ENVI format for which open source image analysis of radial variations for example. Here, the mean processing libraries are available. width of the tangential bands was set to 1mm and the The ENVI files for the 200 discs are provided in the repos - bands were not divided azimuthally. A radial profile of itory under the following path: TreeTrace_spruce/ air-dried density consisting of the average distance to log_end_images/hyperspectral. the pith (meanRadius_mm) and density (airDryDen- sity_kgpm3) of each tangential band is thus obtained for each disc (TreeTrace_spruce/tables/discs_ 2.7.1 Notes on the hyperspectral data density_profiles.txt). Figure  10 provides exemplary images for one wood disc The radial profile of ring widths was coupled with the (#E001B). In the first and second row RGB images taken radial profile of density to assess the mean density of each from the raw and sanded disc surface are shown. In the annual growth ring (airDryDensity_kgpm3). The variable subsequent rows, for each hyperspectral camera, images was added to the disc-ring-level data file. The method is for selected bands are illustrated for the same disc, raw described in Ravoajanahary et al. (2022). and sanded. These disc-ring-level data were used to compute the It can be recognized that the width of the wood disc for data table at the disc level (TreeTrace_spruce/ the FX10 and FX17 images differs. We assume that this is tables/discs_measurements.txt). In this table, caused by the acquisition software and an inaccurate trig- the average density of a disc (airDryDensity_kgpm3) is ger synchronization. Furthermore, a set of cubes got lost the average of all pixels corresponding to wood (exclud- during file storage and are thus missing in the repository. ing bark) in the calibrated CT image. The number of Table  2 shows an overview of the number of captured these pixels was used to compute the under-bark area cubes for each hyperspectral camera. For the FX10 and of the discs (areaUBScanner_cm2). The quadratic mean the raw discs all cubes are available. of the four radii on which ring widths were measured (meanRadiusUBOptical_mm) is also reported. The 2.8 M easurements on wood discs height in the stem of each disc (estimatedHeight_m) was Finally, the 200 discs were transported to INRAE estimated based on the number of rings as explained in Nancy, France, in March 2019, for further analyses. Section 2.1. Ring widths were measured on four orthogonal radii and are provided in the disc-radius-ring-level data file (TreeTrace_spruce/tables/discs_radii_ 3 Access to the data and metadata description rings.txt). From these values, the average profile The images and the database (Longuetaud et  al. 2022b) of annual ring widths of a disc was obtained by sub- are available at Recherche Data Gouv repository: tracting the successive quadratic means of the four https:// doi. org/ 10. 57745/ WKLTJI. Associated metadata radii (i.e. root mean square of external radii of a given access is at https:// metad ata- afs. nancy. inra. fr/ geone ring minus root mean square of internal radii of the twork/ srv/ fre/ catal og. searc h#/ metad ata/ cffee 2f1- 18e1- same ring). The corresponding disc-ring-level data file 4b53- 9f5b- 6cc4c 66f1c b8. (TreeTrace_spruce/tables/discs_rings. The arborescence of the TreeTrace_spruce directory txt) provides cambial age (cambialAge), growth year was detailed in the previous section and is summarized (growthYear), distance to pith (externalRadius_mm) in Fig. 11. and ring width (ringWidth_mm). L onguetaud et al. Annals of Forest Science (2023) 80:9 Page 9 of 12 Fig. 10 Hyperspectral images of cross‑section E001B_le taken with FX10 camera (first row) and FX17 camera (second row) and RGB images taken with Canon device (third row) at Munich The main numeric data are in the TreeTrace_ is accompanied with a variable table (with the same spruce/tables directory. There are five tables, all name followed by _variable) describing each vari- in the form of an array with variables in columns and a able. The data tables are the following with the different first line indicating the variable names. Each data table levels of measurements: Longuetaud et al. Annals of Forest Science (2023) 80:9 Page 10 of 12 Table 2 Number of hyperspectral cubes taken with each • The discs_measurements table contains 200 camera in rough and sanded states at both ends of the 100 lines, one for each disc or log end, and 7 columns. sampled logs • The rgb_images table contains 1164 lines, one for each image on which any measurement has been Hyperspectral Surface Large end Small end camera made, and 8 columns. • The discs_rings table contains 8494 lines, one FX10 Raw 100 100 for each annual growth ring of each disc, and 7 col- FX10 Sanded 88 12 umns. The variables were computed from ring widths FX17 Raw 99 96 measured on orthogonal radii and density profiles FX17 Sanded 78 9 issued from CT images. • The discs_radii_rings table contains 8494 × 4 radii = 33976 lines, one for each ring measurement on each radius, and 6 columns. • The discs_density_profiles table contains 25,230 lines, one for each 1-mm-wide tangential Fig. 11 Organisation of the data files L onguetaud et al. Annals of Forest Science (2023) 80:9 Page 11 of 12 band on which air-dry density was measured on the the site and logs sampled since the trees and logs were CT images, and 5 columns. already cut when we arrived for sampling. We give below some examples of past uses of the database: An automatic reading algorithm of the black and white 4 Technical validation calibration scale was designed by Delconte (2019). All measurements were carefully checked by graphical Decelle and Jalilian (2020) proposed a neural network analyses with R software. The ring measurements were algorithm for automatic segmentation of the wood controlled by using a binocular magnifying glass when in cross-section in the raw images taken in the forest or doubt and by checking for consistency between ring pro- after transportation to Freiburg. files from the four radii measured on each disc. The X-ray The superiority of convolutional neural networks for CT scanner used to measure wood density is controlled the identification of log cross-sections over methods and calibrated by the manufacturer three times a year. adapted from fingerprint or iris recognition techniques A specific calibration procedure (Freyburger et  al. 2009) was demonstrated in the comparative study by Wim- was used to convert Hounsfield Units to wood density. mer et al. (2021a). Convolutional neural networks were also used by 5 Reuse potential and limits Wimmer et  al. (2021b) to identify a log from images All data tables are in plain text format with tabulations as taken by an RGB camera and a CT scanner. separator. They can be imported easily with R or with any Wimmer et  al. (2022) shows that the use of hyper- spreadsheet software. Sometimes information is missing spectral imaging and the combination of several spectra for some variables, which is translated in “NA” value in provides significantly better results than RGB images the tables. alone regarding the visibility of growth rings, which are The RGB images are in standard JPEG or PNG formats. the main characteristic used for traceability. A scale included in each image allows to perform spa- Ravoajanahary et  al. (2022) proposed a method for tial calibration. For a subset of RGB images (one image estimating tree ring density by coupling X-ray CT scan- per log taken at the large end and per site and device), ning and high-resolution optical measurements of tree the black and white scale was manually measured in the ring width. This method was applied to analyse the images and the image resolution is provided in a Table. relationship between ring density and ring width and The CT images of the discs (in the TreeTrace_ cambial age at several height levels in the trees. spruce/log_end_images/xray directory) and logs (in the TreeTrace_spruce/log_images/xray Acknowledgements Thanks to Forêt et Bois de l’Est for helping us by providing the 100 spruce logs directory) can be read with ImageJ software (https:// imagej. and by organizing log transportation to Freiburg. Thanks also to Martin Huber nih. gov/ ij/). from Forest Research Institute of Baden-Württemberg (FVA) and all the team for The registered T-LiDAR clouds (in the TreeTrace_ helping us with the organization of log measurement and imaging at FVA. spruce/log_images/tlidar directory) can be Authors’ contributions visualised with the 3D point cloud processing software FL initiated the funding request; managed the project for the French part; par‑ CloudCompare (https:// www. danie lgm. net/ cc/). ticipated in the data collection, the data analysis and writing the manuscript; and supervised PhD and Master students. RS initiated the funding request, did The hyperspectral images (in the TreeTrace_ his PhD thesis on the basis of these data and participated in the data collec‑ spruce/log_end_images/hyperspectral direc- tion, the data analysis and writing the manuscript. FM initiated the funding tory) require Matlab to be read. request and participated in the data collection, the data analysis and writing the manuscript. TR participated in the data collection, the data analysis and The image database TreeTrace_spruce, providing both writing the manuscript. RD did his PhD thesis on the basis of these data and raw images and ground truth, can be used to develop participated in the data collection, the data analysis and writing the manu‑ and to validate image analysis algorithms for detecting script. TC participated in the data collection, the data analysis and writing the manuscript. PN supervised the PhD and Master students and participated in various features (e.g. pith location, cross-sectional area, the data analysis. IDR supervised the PhD and Master students and partici‑ annual ring widths) on untreated (i.e.  freshly cut) cross- pated in the data analysis. KE supervised the PhD and Master students and sections of logs. participated in the data analysis. AP supervised the PhD and Master students and participated in the data analysis. FB participated in the data collection and The main objectives for the further development of writing the manuscript. AU initiated the funding request, managed the project algorithms are (1) to trace the logs from the forest to the for the Austrian part, participated in the data collection and supervised the sawmill and/or (2) to estimate the wood quality directly PhD and Master students. All authors read and approved the final manuscript. in the forest, on the log yard or on the sawmill line before Funding sawing. Both were objectives of the TreeTrace project. SilvaTech facility is supported by the French National Research Agency The difficulty in such studies performed in an indus - through the Laboratory of Excellence ARBRE (ANR‑11‑LABX ‑0002‑01). This research was made possible thanks to the financial support of the French trial context is that very few information is available on Longuetaud et al. Annals of Forest Science (2023) 80:9 Page 12 of 12 National Research Agency (ANR) and the Austrian Science Fund (FWF) in the Norell K (2011) Automatic counting of annual rings on Pinus sylvestris end framework of the TreeTrace project, ANR‑17‑ CE10‑0016. faces in sawmill industry. Comput Electron Agric 75(2):231–237. https:// doi. org/ 10. 1016/j. compag. 2010. 11. 005 Availability of data and materials Norell K, Borgefors G (2008) Estimation of pith position in untreated log ends The images and the database are available at Research Data Gouv repository: in sawmill environments. Comput Electron Agric 63(2):155–167. https:// https:// doi. org/ 10. 57745/ WKLTJI.doi. org/ 10. 1016/j. compag. 2008. 02. 006 Ravoajanahary T, Mothe F, Longuetaud F (2022) A method for estimating tree Code availability ring density by coupling CT scanning and ring width measurements: Not applicable. Application to the analysis of the ring width ‑ ring density relationship in Picea abies trees. Trees ‑ Structure and Function. https:// doi. org/ 10. 1007/ s00468‑ 022‑ 02373‑2 Declarations Schraml R, Hofbauer H, Petutschnigg A, Uhl A (2015) Tree log identification based on digital cross‑section images of log ends using fingerprint and Ethics approval and consent to participate iris recognition methods. In: Computer Analysis of Images and Patterns. Not applicable. CAIP 2015. Lecture Notes in Computer Science, vol 9256. Springer, pp 752–765. https:// doi. org/ 10. 1007/ 978‑3‑ 319‑ 23192‑1_ 63 Consent for publication Schraml R, Uhl A (2013) Pith estimation on rough log end images using local Not applicable. fourier spectrum analysis. In: Proceedings of the 14th Conference on Computer Graphics and Imaging (CGIM’13), Innsbruck, AUT. https:// doi. Competing interests org/ 10. 2316/P. 2013. 797‑ 012 The authors declare that they have no competing interests. Schraml R, Uhl A (2014) Similarity based crosssec ‑ tion segmentation in rough log end images. In: Artificial Intelligence Applications and Innovations. AIAI 2014. IFIP Advances in Information and Communication Technology, vol Received: 20 October 2022 Accepted: 25 January 2023 436. Springer, pp 614–623. https:// doi. org/ 10. 1007/ 9783‑662‑ 44654‑ 6_‑ 61 Stängle SM, Brüchert F, Heikkila A, Usenius T, Usenius A, Sauter UH (2015) Potentially increased sawmill yield from hardwoods using x‑ray com‑ puted tomography for knot detection. Ann For Sci 72(1):57–65. https:// doi. org/ 10. 1007/ s13595‑ 014‑ 0385‑1 References Wimmer G, Schraml R, Hofbauer H, Petutschnigg A, Uhl A (2021a) Two‑stage BFW (2008) Bundesforschungszentrum für wald. (2008) bfw‑praxisinformation cnn‑based wood log recognition. In: Computational Science and Its nr. 16 ‑ 2008. https:// bfw. ac. at/ 030/ pdf/ 1818_ pi16. pdf . Accessed 11 Jan Applications. ICCSA 2021. Lecture Notes in Computer Science, vol 12955. Springer, pp 115–125. https:// doi. org/ 10. 1007/ 978‑3‑ 030‑ 87007‑2_9 BFW (2022) Bundesforschungszentrum für wald. (2022) Österreichische Wimmer G, Schraml R, Hofbauer H, Petutschnigg A, Uhl A (2022) An analysis waldinventur. periode 2016/21. https:// www. waldi nvent ur. at. Accessed of the use of hyperspectral data for roundwood tracking. In: The 22nd 11 Jan 2023. International Conference on Computational Science and Its Applications Decelle R, Jalilian E (2020) Neural networks for cross‑section segmentation in (ICCSA 2022). Malaga, https:// doi. org/ 10. 1007/ 978‑3‑ 031‑ 10545‑6_ 21 raw images of log ends. In: 2020 IEEE 4th International Conference on Wimmer G, Schraml R, Lamminger L, Petutschnigg A, Uhl A (2021b) Cross‑ Image Processing, Applications and Systems (IPAS). IEEE, pp 131–137. modality wood log tracing. In: 2021 IEEE International Symposium on https:// doi. org/ 10. 1109/ IPAS5 0080. 2020. 93349 60 Multimedia (ISM). IEEE, pp 191–195. https:// doi. org/ 10. 1109/ ISM52 913. Delconte F (2019) Estimation de la qualité de grumes de bois. Master’s thesis, M2 2021. 00038 informatique, Parcours type Apprentissage, Vision, Robotique (AVR), Univer‑ sité de Lorraine, Nancy, France. https:// hal. univlor‑ ra ine. fr/ hal03603‑ 108 Publisher’s Note FCBA (2022) Memento 2022. p 48. https:// www. fcba. fr/ wp‑ conte nt/ uploa ds/ Springer Nature remains neutral with regard to jurisdictional claims in pub‑ 2023/ 01/ Memen to‑ 2022 WEB‑. pdf . Accessed 11 Jan 2023. lished maps and institutional affiliations. Freyburger C, Longuetaud F, Mothe F, Constant T, Leban JM (2009) Measur‑ ing wood density by means of x‑ray computer tomography. Ann For Sci 66(8):804. https:// doi. org/ 10. 1051/ forest/ 20090 71 Kurdthongmee W (2020) A comparative study of the effectiveness of using popular dnn object detection algorithms for pith detection in cross‑ sectional images of parawood. Heliyon 6(2). https:// doi. org/ 10. 1016/j. heliy on. 2020. e03480 Kurdthongmee W, Suwannarat K, Panyuen P, Sae‑Ma N (2018) A fast algorithm to approximate the pith location of rubberwood timber from a normal camera image. In: 15th International Joint Conference on Computer Sci‑ ence and Software Engineering (JCSSE). IEEE, pp 1–6. https:// doi. org/ 10. 1109/ JCSSE. 2018. 84573 75 Longuetaud F, Mothe F, Fournier M, Dlouha J, Santenoise P, Deleuze C Re Read ady y to to submit y submit your our re researc search h ? Choose BMC and benefit fr ? Choose BMC and benefit from om: : (2016) Within‑stem maps of wood density and water content for characterization of species: a case study on three hardwood and two fast, convenient online submission softwood species. Ann For Sci 73(3):601–614. https:// doi. org/ 10. 1007/ s13595‑ 016‑ 0555‑4 thorough peer review by experienced researchers in your field Longuetaud F, Pot G, Mothe F, Barthelemy A, Decelle R, Delconte F, Ge X, rapid publication on acceptance Guillaume G, Mancini T, Ravoajanahary T, Butaud JC, Collet R, Debled‑ support for research data, including large and complex data types Rennesson I, Marcon B, Ngo P, Roux B, Viguier J (2022a) Traceability and quality assessment of Douglas fir (Pseudotsuga menziesii (Mirb.) Franco) • gold Open Access which fosters wider collaboration and increased citations logs: the TreeTrace_Douglas database. Ann For Sci 79:46. https:// doi. org/ maximum visibility for your research: over 100M website views per year 10. 1186/ s13595‑ 022‑ 01163‑7 Longuetaud F, Schraml R, Mothe F, Ravoajanahary T, Decelle R, Constant T, At BMC, research is always in progress. Ngo P, Debled‑Rennesson I, Entacher K, Petutschnigg A, Brüchert F, Uhl A (2022b) “TreeTrace_spruce”. [dataset]. Recherche Data Gouv. Repository, Learn more biomedcentral.com/submissions V1. https:// doi. org/ 10. 57745/ WKLTJI

Journal

Annals of Forest Science – Springer Journals

Published: Feb 13, 2023

Keywords: Image analysis; RGB images; Computed tomography; Picea abies; Wood density; Growth ring width; Terrestrial LIDAR

References

  • Measuring wood density by means of x-ray computer tomography
    Freyburger, C; Longuetaud, F; Mothe, F; Constant, T; Leban, JM
  • Within-stem maps of wood density and water content for characterization of species: a case study on three hardwood and two softwood species
    Longuetaud, F; Mothe, F; Fournier, M; Dlouha, J; Santenoise, P; Deleuze, C
  • Automatic counting of annual rings on Pinus sylvestris end faces in sawmill industry
    Norell, K
  • Estimation of pith position in untreated log ends in sawmill environments
    Norell, K; Borgefors, G
  • A method for estimating tree ring density by coupling CT scanning and ring width measurements: Application to the analysis of the ring width - ring density relationship in Picea abies trees
    Ravoajanahary, T; Mothe, F; Longuetaud, F
  • Potentially increased sawmill yield from hardwoods using x-ray computed tomography for knot detection
    Stängle, SM; Brüchert, F; Heikkila, A; Usenius, T; Usenius, A; Sauter, UH