Image quality assessment based on the space similarity decomposition model

In the image quality assessment (IQA) research field, the structural similarity index measurement (SSIM) method and human visual system (HVS) model have received much attention. However, this paper shows that the definition of the luminance comparison function in SSIM conflicts with the Weber-Fechner law in HVS. To reconcile this contradiction, we propose a space similarity decomposition model based on the Weber-Fechner law and SSIM's structure parameter to decompose the image into mean value, similarity, and vertical components. Experimental results show that the vertical component is the main factor of subjective perception and should be used by itself to assess image quality. This method not only can be used to independently assess image quality, it can also be combined with other IQA methods as a preprocessing step.