No-Reference Image Quality Assessment via Degenerate Decision Logic

In this study, a no-reference (NR) IQA method based on degenerate decision logic (DDL) is proposed. DDL uses wavelet transform to extract the wavelet sub-band features of images and then maps these features to higher- level features through a hybrid network, thereby achieving effective extraction of mixing characteristics and avoiding the redundant information of low level features. DDL also adds decision logic to the assessment, combining decision logic differentiation training and fine-tuning training to allow the assessment result to approach that of human subjective cognition as closely as possible. The performance of the proposed DDL method is evaluated on the LIVE database.