Cycleiqa: Blind Image Quality Assessment Via Cycle-Consistent Adversarial Networks
Peiyun Zhang, Xiao Jun Shao, Zihan Li · 2022 IEEE International Conference on Multimedia and Expo (ICME) · 2022
Existing blind image quality assessment (BIQA) methods aim to extract distortion features by training deep learning models to predict quality scores. However, images suffer from various distortions. Training a single model is typically hard to handle distortion variation problems. To solve the problem, we propose a novel BIQA method to model the quality degradation process caused by image distortions. It consists of a generative adversarial network (GAN)-based quality perception network and a quality regression network. The GAN-based quality perception network is designed to simulate the process of distortion information introduced to images in both forward and reverse directions. The quality regression network extracts the learned hierarchical restoration features from the quality perception network to learn the relationship between features and quality scores. Experimental results on four representation IQA datasets show that the proposed method achieves comparable quality prediction performance with other state-of-the-art methods.