Image quality assessment method based on human visual system
Miao Fen Zhu, Fan Wang, Jinping Ni, Rongli Guo · Optical Sensing and Imaging Technologies and Applications · 2018
Structural Similarity Index (SSIM) is a common and useful image assessment method, and is better than mean square error (MSE) and peak signal to noise ratio (PSNR). However, when evaluating the quality of blur images and noise images, the correlation coefficient between the assessment results of SSIM and the subjective ones is low. A method based on visual structural similarity (VSSIM) for image quality assessment is proposed in this paper. The method is based on the multichannel properties of the log-polar Gabor filter and contrast sensitivity function. Multi-channel visual feature of distorted and reference images with log-polar Gabor transformation is extracted, and then the value of each channel’s SSIM is calculated, and finally, all the SSIM are mixed together according to the weight calculated by the contrast sensitivity function. Experimental results show that VSSIM can assess the blur and noise image quality precisely.