No-reference Image Quality Assessment Algorithm Based on Wavelet Energy and Texture Analysis

Abstract

A new no-reference image quality assessment(QA) algorithm based on wavelet energy and texture analysis (WETA) is proposed in this paper. The detection of common distortion metrics, like block effects, blurring and noise is the basis of WETA algorithm. The wavelet energy difference is used as macroscopic statistical feature to compensate for the limitation of basic distortion detection, which is obtained by natural lossless images energy model. To cover the masking effect of local background and brightness to distortion performance in human eye, this paper utilizes the derivative features of gray level co-occurrence matrix(GLCM) to represent texture feature and image complexity. Finally, the objective quality assessment method is given by fusing distortion performance, wavelet energy difference and texture information into BP neural network for studying. Experiments show that WETA is high consistent to subjective QA scores without referring to original image, and less time-consuming.