Influence of the Smooth Region on the Structural Similarity Index

The Structural Similarity (SSIM) Index is a very popular image quality assessment algorithm due to its good performance and simple calculation. This paper analyzes the influence of the smooth region on the performance of SSIM. It was found out that SSIM tends to depend on the quality of the smooth region to assess the quality of the whole image. From analysis by means of the SSIM quality map, we found out that SSIM overestimated the importance of the smooth region to visual quality. So in this paper we proposed a new weighting function as the spatial pooling method. The resultant weighted-SSIM outperforms the classic SSIM by a considerable amount.

[1]  Alan C. Bovik,et al.  A Statistical Evaluation of Recent Full Reference Image Quality Assessment Algorithms , 2006, IEEE Transactions on Image Processing.

[2]  Stefan Winkler.  Video quality and beyond , 2007, 2007 15th European Signal Processing Conference.

[3]  Eero P. Simoncelli,et al.  Image quality assessment: from error visibility to structural similarity , 2004, IEEE Transactions on Image Processing.

[4]  Zhou Wang,et al.  Spatial Pooling Strategies for Perceptual Image Quality Assessment , 2006, 2006 International Conference on Image Processing.

[5]  Anush Krishna Moorthy.  Spatial Pooling Strategies for Image Quality Assessment , 2010 .

[6]  Zhou Wang,et al.  Stimulus synthesis for efficient evaluation and refinement of perceptual image quality metrics , 2004, IS&T/SPIE Electronic Imaging.