Perceptual Quality Analysis in Deep Domains Using Structure Separation and High-Order Moments
Weizhi Xian,Mingliang Zhou,Bin Fang,Tao Xiang,Weijia Jia,Bin Chen
DOI: https://doi.org/10.1109/tmm.2023.3293730
IF: 7.3
2023-01-01
IEEE Transactions on Multimedia
Abstract:Images are composed of “things” (i.e., structured objects) and “stuff” (i.e., textured surfaces), which have completely different effects on the human visual system (HVS). A good image quality assessment (IQA) method should fully consider the visual salience effects of image structures and the masking effects of image textures. In this paper, we propose a perceptual quality analysis model using structure separation and high-order moments (SSHMPQA) in the deep domain. First, we use a total variation (TV) model to separate the perceptual structures in images from their deep feature maps, thereby maintaining meaningful object shapes with texture suppression and defining perceptual structure-aware distances in the deep domain. Then, we use the first- to fourth-order moments to calculate the mean, skewness and kurtosis of the probability distributions of the deep features. On this basis, we define a perceptual texture-aware distance in the deep domain. We then formulate the final model by solving a well-defined perceptual optimization problem. The proposed SSHMPQA model has good interpretability and is data-driven; moreover, the model does not require a complex and long training process because the optimization problem is convex and has an exact analytical solution. To verify the effectiveness of our model, comprehensive experiments are conducted. The experimental results show that the proposed model is superior to other state-of-the-art traditional and deep learning-based full-reference (FR) IQA methods. The code is publicly available at https://github.com/WeizhiXian/SSHMPQA.
computer science, information systems,telecommunications, software engineering