Towards efficient blind quality evaluation of screen content images based on edge-preserving filter

Abstract

The problem of quality assessment of screen content images from two respects is addressed, the edge-preserving filter based free energy and structural degradation model. For screen content images which always contain texts, edge information plays an important role in the process of evaluating the screen images quality. Inspired by this, the edgepreserving filter based free energy entropy and structural degradation model were combined to extract the quality features and accordingly the statistical model of screen content images was established. Experimental results prove that the proposed method can produce highly consistent with human perception, particularly superior to state-of-the-art full- and no-reference quality metrics on the SIQAD database dedicated to the quality assessment of screen images.