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

Learn More →

A Study of 3D Modeling for Conservation Work of Large-scale Industrial Heritage Structures: Using the South Chimney of Taiwan Tile Corporation′s Takao Factory as a Case Study

A Study of 3D Modeling for Conservation Work of Large-scale Industrial Heritage Structures: Using... Large-scale heritage structures with historical value often suffer damage as a result of external forces. If they are to be rebuilt, while preserving their original appearance, the difficulties of reconstructing the often shattered remains must be addressed. In this paper, the south chimney of the Taiwan Tile Corporation's Takao factory, which was seriously damaged during an earthquake, is used as the example. To develop the integrated technology of 3D laser scanning, with a Fuzzy c-mean algorithm and precision irregular 3D building digital model, eleven complete precision 3D digital models of the south chimney and a 3D digital model of the south chimney needed to be created from 2D architectural survey figures. Judging whether the positions of the undamaged remains match their original positions requires the use of 3D overlay recognition through position, size and appearance. The recognition results are successful for eight pieces of the remains, and the success matching rate is about 72%. The 3D digital recovery simulation model using a precision irregular 3D building digital model can improve the ability to rebuild in keeping with the original appearance, as well as reduce conservation costs in the rebuilding of large-scale heritage structures. Keywords: large-scale heritage structures; 3D laser scanning; Fuzzy c-mean algorithm; 3D building digital model; recovery simulation mode 1. Introduction done primarily via the use of the digital method. As for Large-scale industrial heritage structures, as with smaller pieces and debris too damaged to accurately other historic buildings, can be damaged during identify, replication or the use of new filler material natural disasters, such as earthquakes and severe is recommended. However, in order to generate climate change, bio-deterioration and human-induced a good quality 3D model simulation, the existing damage (e.g., wars or deliberate acts of destruction). remnants must be subjected to 3D laser scanning. However, the conservation of large-scale industrial The obtained 3D point cloud data and high-resolution heritage structures poses numerous difficulties, digital photographs from 3D laser scanning provide mainly due to their massive scale or unusual design. accurate space data, imagery details and other raw data Hence, conventional on-site surveying techniques and necessary for the reconstruction of a 3D simulation 1) mapping-cum-comparison works for the architectural model of the ruined structure . At the same time, the details of these structures cannot be applied. 'pre-damage' version of the simulation model of the Alternatively, the digital modeling approach can be industrial building or structure can be reconstructed considered. With the help of computer simulation with reference to photographs prior to the damage. technology, the reassembly of the ruined parts can be Digital comparison between the 'pre-damage' model tried out beforehand. In this way, the overall efficiency and the 'post-damage' model can then be carried out of the actual restoration or repair work is significantly so as to effectively determine the best conservation increased, reducing time wastage, as well as resulting solution for the damaged structure. in a more accurate restoration. In this paper, the Taiwan Tile Corporation's Takao In order to achieve an efficient and accurate 3D factory was used as a case study for discussion (Fig.1.). simulation, a trial fitting of the larger remnants can be Erected in 1899, the factory was the first ceramic- producing factory established in Kaohsiung City during the Japanese occupation period. With the advent of *Contact Author: Tsung-Chiang Wu, Assistant Professor, industrial technology, the Hoffman kiln was introduced Department of Civil and Engineering Management, into the factory line in 1913 to increase production National Quemoy University, Kinmen, Taiwan efficiency. Being the first of its kind to be introduced Tel: +886-8231-3518 Fax: +886-8231-3528 in Taiwan, this sophisticated technology soon became E-mail: tsung_chiang@nqu.edu.tw a patent product during that period. Having survived ( Received April 1, 2014 ; accepted November 12, 2014 ) Journal of Asian Architecture and Building Engineering/January 2015/158 153 more than a century of industrial development, the 2. Experiment old machineries and facilities could no longer remain In order to achieve an accurate 3D model of the competitive and the entire factory officially closed ruined south chimney, 3D laser scanning technology in 2002. In recognition of its historic, economic and was used to produce accurate measurements and survey 3) technological contributions to Taiwan's history, in 2005 data on-site . An overview of the 3D laser scanner, the entire factory complex, comprising the Hoffman the scanning procedures on-site and the subsequent kiln, the north and south brick chimneys (46 m and 33 construction sequence of the 3D simulation model are 2) m high, respectively , an administrative office and all summarized in the following sections. the related ancillary building structures, were listed as 2.1 Research Tool a National Historic Monument. There are various kinds of commercial 3D laser Unfortunately, on March 4, 2010, the 6.4-magnitude scanners available; however, the different models Chia-hsien earthquake hit in the vicinity of the are only suitable for specific situations or on-site Kaohsiung area and resulted in significant damage requirements. In the case of Taiwan Tile Corporation's to most of the factory buildings. The damage to the Takao factory, factors such as the large scale of the site south chimney was the most serious, as part of the to be surveyed, structural irregularities and the extreme top chimney portion collapsed during the shaking height of the two chimneys had to be considered when (Fig.2.). Emergency rescue work was immediately selecting the scanner. After much deliberation, a FARO begun to investigate the extent of the damage to the Photon 120 3D Laser Scanner, with a Nikon D200 10 south chimney. Due to the height of the chimney mega-pixel digital camera attached on the top of the and possible structural instability, the conventional scanner was selected, as it could perform the scanning surveying method was neither feasible nor advisable; and color photography at the same time. Features such hence, a 3D laser scanner and aerial ladder were as a 360° horizontal field of view, a 320° vertical field used to conduct the on-site survey. In this paper, of view, a scanning speed of 976,000 points/sec with a based on the abovementioned scanning approach, a reach of 153 m and the ability to achieve an accuracy 4) simulated recovery model of the damaged structure rate of ± 2 mm systematical distance error at 25 m was constructed so as to provide a basis for future underlined its suitability for this case study. conservation work. The point cloud data obtained from the laser scanning was first converted to a compatible data format to allow Rapidform©, a 3D reverse engineering software, to construct the model of the ruins. Once the model was constructed, 3D MAX© software was used to generate a simulated version of the constructed model. 2.2 Laser Scanning Process of the Ruined Structure The earthquake caused significant damage to the south chimney, with the top portion of bricks falling and scattering over the ground below. The fallen bricks varied in shape and size, and could be roughly categorized as 'identifiable' and 'unidentifiable'. Big chunks of ruined blocks and individual bricks with different degrees of damage were classed as identifiable, while smaller broken pieces or debris Fig.1. Condition of the Factory Prior to Earthquake Damage were classed as unidentifiable (Fig.3.A). Only the identifiable types were focused on in this study. A total of twenty ruined brick pieces were chosen and subjected to 3D laser scanning (Fig.3.B) to obtain the 3D point cloud data for the subsequent model construction and simulation procedures (Fig.3.C). 2.3 Making of the 3D Ruined Brick Model By using the Rapidform© reverse engineering software, the 3D point cloud data was digitally converted via the MESH function in order to generate the 3D models of the ruined bricks. MESH function of Rapidform© processed the Triangular Irregular Network (TIN) of the point cloud data on the surface of the ruined bricks and computed the normal vector of each triangular mesh. Afterwards, fuzzy c-mean 5),6) algorithm (Samarjit Das, 2013) was applied to Fig.2. Extent of Damage to the South Chimney's Top Portion 154 JAABE vol.14 no.1 January 2015 Tsung-Chiang Wu A C Fig.3. (A) Smaller Broken Pieces or Debris Beyond Identification; (B) Identifiable Ruins Under going Laser Scanning; (C) A Typical 3D Point Cloud Sample View of a Ruined Piece of Brick analyze the positions and states of the ruined bricks. The normal vector of each triangular mesh of the neighboring position was clustered by using the fuzzy c-mean algorithm as the basis for the determination of the range of failure surface. The steps are as follows: first, the number of failure surfaces of a certain ruined brick i is determined to obtain V ¯ starting normal vector. Then, the difference from the normal vector of each triangular mesh X ¯ to the starting normal vectors d( X ¯ , V ¯ ) is determined by Eq. (1). Next, the j i difference from the normal vector of each triangular mesh to the staring normal vector is used to compute membership matrix u as shown in Eq. (2). m value is ij to adjust the fuzzy degree of membership matrix and is greater than 1. When the value is greater than 1, the degree of "fuzzy" membership is higher. In this study, Fig.4. The Range of Failure Surface and Normal Vector 7) the value m=2 by referring to . By using the repeated conventional tedious and labor-intensive method of iteration, the final normal vector and the corresponding d (X ,V ) 1 (X V ) (1) j i j i manually fitting together the pieces of ruined brick in membership matrix are identified. a trial-and-error manner could be avoided. The use of 2 Rapidform© software not only saved time and labor, d (X ,V ) 1 (X V ) (1) j i j i but also generated all kinds of angles and possibilities which might be too physically demanding to attempt, (m1) especially when the object is of considerable weight.   By analyzing digitally the 3D point cloud data of   d (X ,V ) j i the selected twenty pieces of ruined bricks and their   u  (2) ij (m1) respective properties, such as color tone, shape and (m1) i    position on the ground, eleven 'completed' brick ruins     d (X ,V ) j i were successfully restored.   d (X j ,V ) i1 i   u  (2) ij 2.4 Reconstruction of the South Chimney: 3D Model (m1)   As well as the construction of the 3D models of For each ruined brick, the range of failure surface   individual ruined brick pieces, the 3D model of the is confirmed and the normal vectors of each failure d (X j ,V ) i1 i   south chimney (the origin of the bricks) was also surface are obtained. The approximate degree of the required so as to effectively achieve the ultimate area of each failure surface and the normal vector simulated recovery model of the ruins. Based on past azimuth angle difference by 180° are used to judge research reports and drawings, the 3D model of the whether the two failure surfaces are originally south chimney was constructed. By cross-referencing connected (Fig.4.). the 3D cloud data of the south chimney (Fig.6.A) Next, with reference to the ground area coverage and old photographs prior to the earthquake, the of the fallen bricks found on-site, factors such as the measurements of the chimney were counter-checked to effects of the impact when hitting the ground and the ensure an accurate production of the 3D model of the possible damage patterns upon impact were worked south chimney (Figs.6.B & 6.C). out. Based on this information, the ruined pieces of 2.5 Recovery Simulation of the Ruined Chimney brick were carefully compared and analyzed so as to It was found that, as a result of lateral seismic force, enable the correct merging of the broken pieces (Fig.5.). 8) the initial breakage and the falling direction of the With the help of 3D 'trial-fitting' technology , the JAABE vol.14 no.1 January 2015 Tsung-Chiang Wu 155 Fig.5. Trial Merging Process of the Ruined Brick Pieces: (A) Before; (B) After A C Fig.6. (A) Counter-checking the 3D Model with Existing 3D Cloud Data; Completed 'Pre-damage' Version of the South Chimney Model (B) & (C) eleven 'restored' pieces of the south chimney mainly significantly increased. Such was the case for the arose from a single orientation (Fig.7.A). A review of eleven pieces of brick, whereby the breakage locations the exterior surfaces and geometric shapes of each 3D of eight pieces (72% success matching rate) could be model and the overlapping of these 3D ruined models traced successfully with the use of the above method. with the reconstructed 3D pre-damage version of The conventional repair approach usually involves the south chimney model allowed the position of the conservators on-site who manually conduct trial-and- ruined models to be traced (Figs.7.B & 7.C). Running error fittings repeatedly until the correct pieces are the recovery model simulation allowed eight of the found and matched. But when the ruined pieces are eleven 'restored' ruined pieces to be accurately traced; too bulky to handle, manual comparison becomes suitable matching positions for only three of the ruined physically impossible. When the proposed 3D model pieces could not be found (Fig.8.). simulation approach is used, the conventional tedious 2.6 Results and Analysis a nd t i m e -c onsum i ng m et hod c an be avoi de d. By With the help of 3D laser scanning and high- digitally comparing the 'pre-damage' condition of the resolution digital imaging technology, the 3D chimney and re-confirming the ruined pieces with coordinate and color data thus obtained helped in the actual on-site damage positions, the drafting of the creation of more accurate 3D models for the restoration plan procedures could be carried out more subsequent reconstruction simulation of the brick ruins. effectively and efficiently. The proposed approach The accuracy of the south chimney model, constructed reduced the risks and financial burden of restoration, mainly from the historical survey information, was and the digital information obtained has provided a further enhanced with 3D point cloud data, so as to basis for future conservation works. achieve a near-accurate and realistic representation of the extent of the on-site damage. 3. Conclusions By overlapping the 3D models of the ruined bricks The practice of building complete and accurate 3D and the 3D south chimney model, and simultaneously spatial models for high-risk (e.g., natural disasters cross-referencing the exterior surface and geometric and war) historical buildings and sites should be shape properties of the individual brick pieces, the recommended to serve as a background reference success matching rate for the ruined bricks could be for any repair of historic structures in the future. By 156 JAABE vol.14 no.1 January 2015 Tsung-Chiang Wu C Fig.7. (A) Distribution Pattern of the Ruined Brick Pieces Around the South Chimney; (B) & (C) Proposed Positions of the Ruined Bricks Fig.8. (A) & (B) Recovery Simulation Model Showing the Proposed Positioning of the Eight Ruined Pieces on the South Chimney; (C) Aerial View 3) M. Brizzi, S. Court, A. d'Andrea, A. Lastra3 and D. Sepio (2006) combining the 3D laser scanning technology and Fuzzy "3D laser scanning as a tool for conservation: The Experiences c-mean algorithm, the 3D digital model of the ruined of the Herculaneum Conservation Project" The 7th International bricks can be reconstructed. This approach could Symposium on Virtual Reality, Archaeology and Cultural Heritage generate models with complete exterior appearance VAST. and precise geometrical information. It would also 4) http://www.faro.com/focus/uk 5) Samarjit Das (2013) "Pattern Recognition using the Fuzzy provide a basis for reference on pre-damage conditions c-means Technique" International Journal of Energy, Information and post-damage repairs. The success of the computer and Communications, Vol. 4, pp.1-14. model simulation method in working out the recovery 6) Zheng-Yi Feng, Yi-Kai Chen, Chang-Hai Chien* (2010) model for the ruined brick chimney has proven its "Development of Attitude Classification Program for Rock Slope feasibility for future conservation projects of large- Surfaces" Journal of Soil and Water Conservation, Vol. 42(2), pp.167-176. scale historic buildings and sites. 7) Hammah, R. E. and Curran, J. H., (1998) "Fuzzy cluster algorithm for the automatic identification of joints sets", Int. J. Rock. Mech. Acknowledgment Min., Vol. 7, pp.889-905. This research was supported by the Taiwan National 8) G. Carra, S. D'Amelio, B. Villa (2008) "The virtual construction of Science Council: "A simulation of historic building temple B in Selinunte excavation site" The International Archives of the Photogrammetry, Remote Sensing and Spatial Information restoration using integration of digital technology Sciences. Vol. XXXVII. Part B5. Beijing. and fuzzy clustering theory, case study: Dagou 9) Zhou Keqin, Zhao Xu, Zhou Junzhao, Wang Fei, Hu Song Manufactory, Taiwan Tile Corporation" (Project No. (2008) "Application of terrestrial laser scanning for heritage 100-2410-H-507-009-). conservation in Yungang Grotto", The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. XXXVII. Part B5. pp.337-340, Beijing. References 10) Marcello Balzani, Nicola Santopuoli, Alessandro Grieco, Nicola 1) http://rapidform.zendesk.com/home Zaltron (2004) "Laser Scanner 3D Survey in Archaeological Field: 2) NCKU Research Development Foundation: The Research and the Forum Of Pompeii", International Conference on Remote Investigative Report on Kaohsiung city listed building – Taiwan Sensing Archaeology, pp.169-175, Beijing. Brick and Tile Production Company Kaohsiung factory (Chong- 11) R. Brumana, L. Fregonese, F. Fassi, F. De Pascalis, (2005) "3D Du Tang-Rong Brick Kiln Factory), Taiwan, Kaohsiung City Laser scanner points clouds and 2D multi-spectral images: a data Government Cultural Heritage Bureau, 2005. (in Chinese). matching software for cultural heritage conservation" CIPA 2005 XX International Symposium, 26 September – 01 October, 2005, Torino, Italy. JAABE vol.14 no.1 January 2015 Tsung-Chiang Wu 157 12) Naai-Jung Shih, Huey-Jiun Wang, Chen-Yan Lin, and Chai-Yuan Liau, "3D scan for the digital preservation of a historical temple in Taiwan," Journal of Advances in Engineering Software, Vol. 38, Is. 7, pp.501-512, July 2007. 13) J.L., Architect Studio: The Urgent Support Protection Project Report on Kaohsiung city listed building – Taiwan Brick and Tile Production Company Kaohsiung factory (Chong-Du Tang- Rong Brick Kiln Factory), Taiwan, Kaohsiung City Government Cultural Heritage Bureau, 2007. (in Chinese). 14) P. Salonia, V. Bellucci, S. Scolastic, A. Marcolong, T. Leti Messina (2007) "3D survey technologies for reconstruction, analysis and diagnosis in the conservation process of cultural heritage" XXI International CIPA Symposium, 01-06 October 2007, Athens, Greece. 15) H. Pires, P. Ortiz, P. Marques, H. Sanchez (2006) "Close- range Laser Scanning Applied to Archaeological Artifacts Documentation" Virtual Reconstruction of an XVIth Century Ceramic Po, The 7th International Symposium on Virtual Reality, Archaeology and Cultural Heritage VAST. 16) Wu, Tsung-Chiang, Yi-Chun Lin, Min-Fu Hsu (2004) "A study of mapping Taiwan historic building using 3D laser scanning technology" Journal of Surveying Engineering, Vol. 46, pp.77-94. 158 JAABE vol.14 no.1 January 2015 Tsung-Chiang Wu http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Asian Architecture and Building Engineering Taylor & Francis

A Study of 3D Modeling for Conservation Work of Large-scale Industrial Heritage Structures: Using the South Chimney of Taiwan Tile Corporation′s Takao Factory as a Case Study

Download PDF
6 pages

Loading next page...
 
/lp/taylor-francis/a-study-of-3d-modeling-for-conservation-work-of-large-scale-industrial-bbr8hexnmP

References (14)

Publisher
Taylor & Francis
Copyright
© 2018 Architectural Institute of Japan
ISSN
1347-2852
eISSN
1346-7581
DOI
10.3130/jaabe.14.153
Publisher site
See Article on Publisher Site

Abstract

Large-scale heritage structures with historical value often suffer damage as a result of external forces. If they are to be rebuilt, while preserving their original appearance, the difficulties of reconstructing the often shattered remains must be addressed. In this paper, the south chimney of the Taiwan Tile Corporation's Takao factory, which was seriously damaged during an earthquake, is used as the example. To develop the integrated technology of 3D laser scanning, with a Fuzzy c-mean algorithm and precision irregular 3D building digital model, eleven complete precision 3D digital models of the south chimney and a 3D digital model of the south chimney needed to be created from 2D architectural survey figures. Judging whether the positions of the undamaged remains match their original positions requires the use of 3D overlay recognition through position, size and appearance. The recognition results are successful for eight pieces of the remains, and the success matching rate is about 72%. The 3D digital recovery simulation model using a precision irregular 3D building digital model can improve the ability to rebuild in keeping with the original appearance, as well as reduce conservation costs in the rebuilding of large-scale heritage structures. Keywords: large-scale heritage structures; 3D laser scanning; Fuzzy c-mean algorithm; 3D building digital model; recovery simulation mode 1. Introduction done primarily via the use of the digital method. As for Large-scale industrial heritage structures, as with smaller pieces and debris too damaged to accurately other historic buildings, can be damaged during identify, replication or the use of new filler material natural disasters, such as earthquakes and severe is recommended. However, in order to generate climate change, bio-deterioration and human-induced a good quality 3D model simulation, the existing damage (e.g., wars or deliberate acts of destruction). remnants must be subjected to 3D laser scanning. However, the conservation of large-scale industrial The obtained 3D point cloud data and high-resolution heritage structures poses numerous difficulties, digital photographs from 3D laser scanning provide mainly due to their massive scale or unusual design. accurate space data, imagery details and other raw data Hence, conventional on-site surveying techniques and necessary for the reconstruction of a 3D simulation 1) mapping-cum-comparison works for the architectural model of the ruined structure . At the same time, the details of these structures cannot be applied. 'pre-damage' version of the simulation model of the Alternatively, the digital modeling approach can be industrial building or structure can be reconstructed considered. With the help of computer simulation with reference to photographs prior to the damage. technology, the reassembly of the ruined parts can be Digital comparison between the 'pre-damage' model tried out beforehand. In this way, the overall efficiency and the 'post-damage' model can then be carried out of the actual restoration or repair work is significantly so as to effectively determine the best conservation increased, reducing time wastage, as well as resulting solution for the damaged structure. in a more accurate restoration. In this paper, the Taiwan Tile Corporation's Takao In order to achieve an efficient and accurate 3D factory was used as a case study for discussion (Fig.1.). simulation, a trial fitting of the larger remnants can be Erected in 1899, the factory was the first ceramic- producing factory established in Kaohsiung City during the Japanese occupation period. With the advent of *Contact Author: Tsung-Chiang Wu, Assistant Professor, industrial technology, the Hoffman kiln was introduced Department of Civil and Engineering Management, into the factory line in 1913 to increase production National Quemoy University, Kinmen, Taiwan efficiency. Being the first of its kind to be introduced Tel: +886-8231-3518 Fax: +886-8231-3528 in Taiwan, this sophisticated technology soon became E-mail: tsung_chiang@nqu.edu.tw a patent product during that period. Having survived ( Received April 1, 2014 ; accepted November 12, 2014 ) Journal of Asian Architecture and Building Engineering/January 2015/158 153 more than a century of industrial development, the 2. Experiment old machineries and facilities could no longer remain In order to achieve an accurate 3D model of the competitive and the entire factory officially closed ruined south chimney, 3D laser scanning technology in 2002. In recognition of its historic, economic and was used to produce accurate measurements and survey 3) technological contributions to Taiwan's history, in 2005 data on-site . An overview of the 3D laser scanner, the entire factory complex, comprising the Hoffman the scanning procedures on-site and the subsequent kiln, the north and south brick chimneys (46 m and 33 construction sequence of the 3D simulation model are 2) m high, respectively , an administrative office and all summarized in the following sections. the related ancillary building structures, were listed as 2.1 Research Tool a National Historic Monument. There are various kinds of commercial 3D laser Unfortunately, on March 4, 2010, the 6.4-magnitude scanners available; however, the different models Chia-hsien earthquake hit in the vicinity of the are only suitable for specific situations or on-site Kaohsiung area and resulted in significant damage requirements. In the case of Taiwan Tile Corporation's to most of the factory buildings. The damage to the Takao factory, factors such as the large scale of the site south chimney was the most serious, as part of the to be surveyed, structural irregularities and the extreme top chimney portion collapsed during the shaking height of the two chimneys had to be considered when (Fig.2.). Emergency rescue work was immediately selecting the scanner. After much deliberation, a FARO begun to investigate the extent of the damage to the Photon 120 3D Laser Scanner, with a Nikon D200 10 south chimney. Due to the height of the chimney mega-pixel digital camera attached on the top of the and possible structural instability, the conventional scanner was selected, as it could perform the scanning surveying method was neither feasible nor advisable; and color photography at the same time. Features such hence, a 3D laser scanner and aerial ladder were as a 360° horizontal field of view, a 320° vertical field used to conduct the on-site survey. In this paper, of view, a scanning speed of 976,000 points/sec with a based on the abovementioned scanning approach, a reach of 153 m and the ability to achieve an accuracy 4) simulated recovery model of the damaged structure rate of ± 2 mm systematical distance error at 25 m was constructed so as to provide a basis for future underlined its suitability for this case study. conservation work. The point cloud data obtained from the laser scanning was first converted to a compatible data format to allow Rapidform©, a 3D reverse engineering software, to construct the model of the ruins. Once the model was constructed, 3D MAX© software was used to generate a simulated version of the constructed model. 2.2 Laser Scanning Process of the Ruined Structure The earthquake caused significant damage to the south chimney, with the top portion of bricks falling and scattering over the ground below. The fallen bricks varied in shape and size, and could be roughly categorized as 'identifiable' and 'unidentifiable'. Big chunks of ruined blocks and individual bricks with different degrees of damage were classed as identifiable, while smaller broken pieces or debris Fig.1. Condition of the Factory Prior to Earthquake Damage were classed as unidentifiable (Fig.3.A). Only the identifiable types were focused on in this study. A total of twenty ruined brick pieces were chosen and subjected to 3D laser scanning (Fig.3.B) to obtain the 3D point cloud data for the subsequent model construction and simulation procedures (Fig.3.C). 2.3 Making of the 3D Ruined Brick Model By using the Rapidform© reverse engineering software, the 3D point cloud data was digitally converted via the MESH function in order to generate the 3D models of the ruined bricks. MESH function of Rapidform© processed the Triangular Irregular Network (TIN) of the point cloud data on the surface of the ruined bricks and computed the normal vector of each triangular mesh. Afterwards, fuzzy c-mean 5),6) algorithm (Samarjit Das, 2013) was applied to Fig.2. Extent of Damage to the South Chimney's Top Portion 154 JAABE vol.14 no.1 January 2015 Tsung-Chiang Wu A C Fig.3. (A) Smaller Broken Pieces or Debris Beyond Identification; (B) Identifiable Ruins Under going Laser Scanning; (C) A Typical 3D Point Cloud Sample View of a Ruined Piece of Brick analyze the positions and states of the ruined bricks. The normal vector of each triangular mesh of the neighboring position was clustered by using the fuzzy c-mean algorithm as the basis for the determination of the range of failure surface. The steps are as follows: first, the number of failure surfaces of a certain ruined brick i is determined to obtain V ¯ starting normal vector. Then, the difference from the normal vector of each triangular mesh X ¯ to the starting normal vectors d( X ¯ , V ¯ ) is determined by Eq. (1). Next, the j i difference from the normal vector of each triangular mesh to the staring normal vector is used to compute membership matrix u as shown in Eq. (2). m value is ij to adjust the fuzzy degree of membership matrix and is greater than 1. When the value is greater than 1, the degree of "fuzzy" membership is higher. In this study, Fig.4. The Range of Failure Surface and Normal Vector 7) the value m=2 by referring to . By using the repeated conventional tedious and labor-intensive method of iteration, the final normal vector and the corresponding d (X ,V ) 1 (X V ) (1) j i j i manually fitting together the pieces of ruined brick in membership matrix are identified. a trial-and-error manner could be avoided. The use of 2 Rapidform© software not only saved time and labor, d (X ,V ) 1 (X V ) (1) j i j i but also generated all kinds of angles and possibilities which might be too physically demanding to attempt, (m1) especially when the object is of considerable weight.   By analyzing digitally the 3D point cloud data of   d (X ,V ) j i the selected twenty pieces of ruined bricks and their   u  (2) ij (m1) respective properties, such as color tone, shape and (m1) i    position on the ground, eleven 'completed' brick ruins     d (X ,V ) j i were successfully restored.   d (X j ,V ) i1 i   u  (2) ij 2.4 Reconstruction of the South Chimney: 3D Model (m1)   As well as the construction of the 3D models of For each ruined brick, the range of failure surface   individual ruined brick pieces, the 3D model of the is confirmed and the normal vectors of each failure d (X j ,V ) i1 i   south chimney (the origin of the bricks) was also surface are obtained. The approximate degree of the required so as to effectively achieve the ultimate area of each failure surface and the normal vector simulated recovery model of the ruins. Based on past azimuth angle difference by 180° are used to judge research reports and drawings, the 3D model of the whether the two failure surfaces are originally south chimney was constructed. By cross-referencing connected (Fig.4.). the 3D cloud data of the south chimney (Fig.6.A) Next, with reference to the ground area coverage and old photographs prior to the earthquake, the of the fallen bricks found on-site, factors such as the measurements of the chimney were counter-checked to effects of the impact when hitting the ground and the ensure an accurate production of the 3D model of the possible damage patterns upon impact were worked south chimney (Figs.6.B & 6.C). out. Based on this information, the ruined pieces of 2.5 Recovery Simulation of the Ruined Chimney brick were carefully compared and analyzed so as to It was found that, as a result of lateral seismic force, enable the correct merging of the broken pieces (Fig.5.). 8) the initial breakage and the falling direction of the With the help of 3D 'trial-fitting' technology , the JAABE vol.14 no.1 January 2015 Tsung-Chiang Wu 155 Fig.5. Trial Merging Process of the Ruined Brick Pieces: (A) Before; (B) After A C Fig.6. (A) Counter-checking the 3D Model with Existing 3D Cloud Data; Completed 'Pre-damage' Version of the South Chimney Model (B) & (C) eleven 'restored' pieces of the south chimney mainly significantly increased. Such was the case for the arose from a single orientation (Fig.7.A). A review of eleven pieces of brick, whereby the breakage locations the exterior surfaces and geometric shapes of each 3D of eight pieces (72% success matching rate) could be model and the overlapping of these 3D ruined models traced successfully with the use of the above method. with the reconstructed 3D pre-damage version of The conventional repair approach usually involves the south chimney model allowed the position of the conservators on-site who manually conduct trial-and- ruined models to be traced (Figs.7.B & 7.C). Running error fittings repeatedly until the correct pieces are the recovery model simulation allowed eight of the found and matched. But when the ruined pieces are eleven 'restored' ruined pieces to be accurately traced; too bulky to handle, manual comparison becomes suitable matching positions for only three of the ruined physically impossible. When the proposed 3D model pieces could not be found (Fig.8.). simulation approach is used, the conventional tedious 2.6 Results and Analysis a nd t i m e -c onsum i ng m et hod c an be avoi de d. By With the help of 3D laser scanning and high- digitally comparing the 'pre-damage' condition of the resolution digital imaging technology, the 3D chimney and re-confirming the ruined pieces with coordinate and color data thus obtained helped in the actual on-site damage positions, the drafting of the creation of more accurate 3D models for the restoration plan procedures could be carried out more subsequent reconstruction simulation of the brick ruins. effectively and efficiently. The proposed approach The accuracy of the south chimney model, constructed reduced the risks and financial burden of restoration, mainly from the historical survey information, was and the digital information obtained has provided a further enhanced with 3D point cloud data, so as to basis for future conservation works. achieve a near-accurate and realistic representation of the extent of the on-site damage. 3. Conclusions By overlapping the 3D models of the ruined bricks The practice of building complete and accurate 3D and the 3D south chimney model, and simultaneously spatial models for high-risk (e.g., natural disasters cross-referencing the exterior surface and geometric and war) historical buildings and sites should be shape properties of the individual brick pieces, the recommended to serve as a background reference success matching rate for the ruined bricks could be for any repair of historic structures in the future. By 156 JAABE vol.14 no.1 January 2015 Tsung-Chiang Wu C Fig.7. (A) Distribution Pattern of the Ruined Brick Pieces Around the South Chimney; (B) & (C) Proposed Positions of the Ruined Bricks Fig.8. (A) & (B) Recovery Simulation Model Showing the Proposed Positioning of the Eight Ruined Pieces on the South Chimney; (C) Aerial View 3) M. Brizzi, S. Court, A. d'Andrea, A. Lastra3 and D. Sepio (2006) combining the 3D laser scanning technology and Fuzzy "3D laser scanning as a tool for conservation: The Experiences c-mean algorithm, the 3D digital model of the ruined of the Herculaneum Conservation Project" The 7th International bricks can be reconstructed. This approach could Symposium on Virtual Reality, Archaeology and Cultural Heritage generate models with complete exterior appearance VAST. and precise geometrical information. It would also 4) http://www.faro.com/focus/uk 5) Samarjit Das (2013) "Pattern Recognition using the Fuzzy provide a basis for reference on pre-damage conditions c-means Technique" International Journal of Energy, Information and post-damage repairs. The success of the computer and Communications, Vol. 4, pp.1-14. model simulation method in working out the recovery 6) Zheng-Yi Feng, Yi-Kai Chen, Chang-Hai Chien* (2010) model for the ruined brick chimney has proven its "Development of Attitude Classification Program for Rock Slope feasibility for future conservation projects of large- Surfaces" Journal of Soil and Water Conservation, Vol. 42(2), pp.167-176. scale historic buildings and sites. 7) Hammah, R. E. and Curran, J. H., (1998) "Fuzzy cluster algorithm for the automatic identification of joints sets", Int. J. Rock. Mech. Acknowledgment Min., Vol. 7, pp.889-905. This research was supported by the Taiwan National 8) G. Carra, S. D'Amelio, B. Villa (2008) "The virtual construction of Science Council: "A simulation of historic building temple B in Selinunte excavation site" The International Archives of the Photogrammetry, Remote Sensing and Spatial Information restoration using integration of digital technology Sciences. Vol. XXXVII. Part B5. Beijing. and fuzzy clustering theory, case study: Dagou 9) Zhou Keqin, Zhao Xu, Zhou Junzhao, Wang Fei, Hu Song Manufactory, Taiwan Tile Corporation" (Project No. (2008) "Application of terrestrial laser scanning for heritage 100-2410-H-507-009-). conservation in Yungang Grotto", The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. XXXVII. Part B5. pp.337-340, Beijing. References 10) Marcello Balzani, Nicola Santopuoli, Alessandro Grieco, Nicola 1) http://rapidform.zendesk.com/home Zaltron (2004) "Laser Scanner 3D Survey in Archaeological Field: 2) NCKU Research Development Foundation: The Research and the Forum Of Pompeii", International Conference on Remote Investigative Report on Kaohsiung city listed building – Taiwan Sensing Archaeology, pp.169-175, Beijing. Brick and Tile Production Company Kaohsiung factory (Chong- 11) R. Brumana, L. Fregonese, F. Fassi, F. De Pascalis, (2005) "3D Du Tang-Rong Brick Kiln Factory), Taiwan, Kaohsiung City Laser scanner points clouds and 2D multi-spectral images: a data Government Cultural Heritage Bureau, 2005. (in Chinese). matching software for cultural heritage conservation" CIPA 2005 XX International Symposium, 26 September – 01 October, 2005, Torino, Italy. JAABE vol.14 no.1 January 2015 Tsung-Chiang Wu 157 12) Naai-Jung Shih, Huey-Jiun Wang, Chen-Yan Lin, and Chai-Yuan Liau, "3D scan for the digital preservation of a historical temple in Taiwan," Journal of Advances in Engineering Software, Vol. 38, Is. 7, pp.501-512, July 2007. 13) J.L., Architect Studio: The Urgent Support Protection Project Report on Kaohsiung city listed building – Taiwan Brick and Tile Production Company Kaohsiung factory (Chong-Du Tang- Rong Brick Kiln Factory), Taiwan, Kaohsiung City Government Cultural Heritage Bureau, 2007. (in Chinese). 14) P. Salonia, V. Bellucci, S. Scolastic, A. Marcolong, T. Leti Messina (2007) "3D survey technologies for reconstruction, analysis and diagnosis in the conservation process of cultural heritage" XXI International CIPA Symposium, 01-06 October 2007, Athens, Greece. 15) H. Pires, P. Ortiz, P. Marques, H. Sanchez (2006) "Close- range Laser Scanning Applied to Archaeological Artifacts Documentation" Virtual Reconstruction of an XVIth Century Ceramic Po, The 7th International Symposium on Virtual Reality, Archaeology and Cultural Heritage VAST. 16) Wu, Tsung-Chiang, Yi-Chun Lin, Min-Fu Hsu (2004) "A study of mapping Taiwan historic building using 3D laser scanning technology" Journal of Surveying Engineering, Vol. 46, pp.77-94. 158 JAABE vol.14 no.1 January 2015 Tsung-Chiang Wu

Journal

Journal of Asian Architecture and Building Engineering – Taylor & Francis

Published: Jan 1, 2015

Keywords: large-scale heritage structures; 3D laser scanning; Fuzzy c-mean algorithm; 3D building digital model; recovery simulation mode

There are no references for this article.