Design and Research of Full Reference Image Quality Assessment Algorithm

Visual signals play an important medium role in human communication with the outside world. However, the various links of the process of image acquisition, storage, transmission all may produce changes to the image information and degradation of the image structure. Therefore, it becomes very important to assess the image quality. However, subjective assessment wastes time and energy, which is not feasible in practical applications, and subjective experiments are affected by many factors such as viewing distance, display equipment, lighting conditions, visual ability of the observer, and emotion. Therefore, it is necessary to design mathematical model that can automatically and accurately predict the subjective quality. This research adopted LOG-PSNR image quality prediction model based on the LOG operator. The model is compared with SSIM and NLOG-MSE two models on the LIVE database. The experimental results suggest that the LOG-PSNR algorithm has good consistency with the subjective scores of the database, it shows that the perceived quality of the image can be well reflected by calculating the difference between the reference image and the distorted image after removing the first-order and second-order statistical redundancy of the image by using the LOG operator.