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

Learn More →

Visual Quality Assessment by Machine LearningMetrics Fusion

Visual Quality Assessment by Machine Learning: Metrics Fusion [There have been an abundance of developed image quality assessment (IQA) metrics during the last decade. However, these is not an individual one, whose performance always tops the performance ranking list on all subjective databases and for all distortions. The combination of multiple IQA metrics is expected to be better than each of them individually used. Two metric fusion frameworks are introduced in this chapter. The one introduces a multi-method fusion (MMF), which gives a MMF score by using a nonlinear combination of the scores computed from the multiple IQA metrics for fusion. The combination is achieved by training a set of weights using support vector regression. The other one presents an ensemble-based framework. In this framework, some features are extracted from the existing IQA metrics. These features are trained to be as basic image quality scorers (BIQSs). For copying with specific distortion types, some advanced features are trained to be as advanced image quality scorers (AIQSs). In addition, two statistical testing methods are employed to do scorer selection. Finally, a machine learning approach is adopted as a score fuser to combine all outputs from the selected scorers.] http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png

Loading next page...
 
/lp/springer-journals/visual-quality-assessment-by-machine-learning-metrics-fusion-n6UoXZ4axs
Publisher
Springer Singapore
Copyright
© The Author(s) 2015
ISBN
978-981-287-467-2
Pages
93 –122
DOI
10.1007/978-981-287-468-9_5
Publisher site
See Chapter on Publisher Site

Abstract

[There have been an abundance of developed image quality assessment (IQA) metrics during the last decade. However, these is not an individual one, whose performance always tops the performance ranking list on all subjective databases and for all distortions. The combination of multiple IQA metrics is expected to be better than each of them individually used. Two metric fusion frameworks are introduced in this chapter. The one introduces a multi-method fusion (MMF), which gives a MMF score by using a nonlinear combination of the scores computed from the multiple IQA metrics for fusion. The combination is achieved by training a set of weights using support vector regression. The other one presents an ensemble-based framework. In this framework, some features are extracted from the existing IQA metrics. These features are trained to be as basic image quality scorers (BIQSs). For copying with specific distortion types, some advanced features are trained to be as advanced image quality scorers (AIQSs). In addition, two statistical testing methods are employed to do scorer selection. Finally, a machine learning approach is adopted as a score fuser to combine all outputs from the selected scorers.]

Published: May 10, 2015

Keywords: Machine learning; Image quality scorer; Multi-method fusion (MMF); Content-dependent MMF (CD-MMF); Context-free MMF (CF-MMF); ParaBoosting ensemble (PBE)

There are no references for this article.