Reduced-reference image quality assessment using moment method. Issue 10 (2nd October 2016)

• Record Type: Journal Article
• Authors: Yang, Diwei; Shen, Yuantong; Shen, Yongluo; Li, Hongwei
• Is Part Of: International journal of electronics. Volume 103:Issue 10(2016)
• Page Start: 1607
• Page End: 1616
• Publication Date: 2016-10-02
• Subjects: Imaging systems -- moment method -- natural scene statistics -- reduced-reference quality assessment -- wavelet transform; Electronics -- Periodicals; Electronic control -- Periodicals; 621.381
• DOI: 10.1080/00207217.2016.1138517
• Languages: English
• ISSNs: 0020-7217

Abstract

ABSTRACT: Reduced-reference image quality assessment (RR IQA) aims to evaluate the perceptual quality of a distorted image through partial information of the corresponding reference image. In this paper, a novel RR IQA metric is proposed by using the moment method. We claim that the first and second moments of wavelet coefficients of natural images can have approximate and regular change that are disturbed by different types of distortions, and that this disturbance can be relevant to human perceptions of quality. We measure the difference of these statistical parameters between reference and distorted image to predict the visual quality degradation. The introduced IQA metric is suitable for implementation and has relatively low computational complexity. The experimental results on Laboratory for Image and Video Engineering (LIVE) and Tampere Image Database (TID) image databases indicate that the proposed metric has a good predictive performance.