HVS-based quality assessment metrics for 3-D images

Abstract

Significant efforts from both the academia and industry have been devoted to advance three-dimensional (3-D) imaging technologies. However, accurate and easy-to-use visual quality assessment metrics still lack for 3-D images. In this paper, we propose three objective quality assessment metrics for 3-D images taking into consideration a set of relevant visual characteristic factors, including contrast sensitivity, multichannel and binocular parallax characteristics. The human visual signal-to-noise ratio (HVSNR), parallax distortion ratio (PDR), and different peak signal-to-noise ratio (DPSNR) are the three independent objective quality assessment metrics proposed in this paper. Experimental results show that the quality assessment results based upon the proposed metrics are consistent with the quality grades obtained by subjective assessment.