Image quality assessment measure based on natural image statistics in the Tetrolet domain

摘要

This paper deals with a reduced reference (RR) image quality measure based onnatural image statistics modeling. For this purpose, Tetrolet transform is usedsince it provides a convenient way to capture local geometric structures. Thistransform is applied to both reference and distorted images. Then, GaussianScale Mixture (GSM) is proposed to model subbands in order to take accountstatistical dependencies between tetrolet coefficients. In order to quantifythe visual degradation, a measure based on Kullback Leibler Divergence (KLD) isprovided. The proposed measure was tested on the Cornell VCL A-57 dataset andcompared with other measures according to FR-TV1 VQEG framework.