Abstract:

Although the Structural Similarity （SSIM） method, a structural information-based method for image quality assessment, performs well with low computational complexity, it is still of some disadvantages because it simply takes the average SSIM of sub-blocks as the mean SSIM （MSSIM） of the whole image without considering the difference of human visual sensitivity in different image areas. In order to solve this problem, this palper proposes a block classification-based image quality assessment method named Weighted Structural Similarity （WSSIM）. In this method, an image is separated into some blocks that are further divided into edge blocks, detail blocks and smooth blocks, and the SSIMs of different types of blocks are weighted with different values to calculate the WSSIM of the whole image. Experimental results indicate that the proposed method is superior to the MSSIM method and the variance-based weighted SSIM.

Key words: image quality, assessment, structural similarity, block classification, weighted average, human visual system