Blind Image Quality Assessment Based on Mutual Information

Abstract

Mutual information can indicate the degradation severity of natural images by capturing images' local structural properties. In this paper, we analyze the dependence between neighboring pixels and propose a no reference image quality assessment method based on mutual information in wavelet domain. The proposed image quality assessment (IQA) method, named MIQA-II, computes mutual information between neighboring pixels in the same subband, across different orientations and scales as features. A quadratic function is used to fit the mutual information values along a distance and the features are then transformed to a final predicted quality score through a two-step framework. MIQA-II is tested on the LIVE IQA database. The experimental results show that MIQA-II has a competitive subjective relevance performance and acceptable time consumption. In addition, MIQA-II still achieves good performance with a small training set.

Improving the Performance of No-Reference Image Quality Assessment Algorithm for Contrast-Distorted Images Using Natural Scene Statistics

“…In [19], Liu et al proposed a no-reference image quality assessment method depending on mutual information of wavelet domain named (MIQA-II), by computing neighbouring pixels using LIVE [20] database. This method showed good results in identifying distortion.…”