Reduced-Reference 3D Image Quality Measurement via Spatial to Gradient Domain Feature Aggregation
Jian Ma, Xiyu Han, Guoming Xu · 2021
Objective quality measurement of a three-dimensional (3D) image is a challenging issue in various 3D visual applications. In this paper, we proposes a novel reduced-reference (RR) 3D quality assessment evaluator to deal with the characteristics of 3D images. Specifically, in spatial domain, the generalized Gaussian density (GGD) fits of luminance wavelet coefficients, and the correlations of luminance and disparity wavelet coefficients are used to represent the statistical characteristics of 3D image. Furthermore, in gradient domain, the enhanced gradient magnitudes are computed by using neighborhood phase congruency (PC) information to weight the gradient magnitudes (GM) in a locally adaptive manner. Afterward, the entropy differencing of discrete wavelet transform coefficients of enhanced gradient magnitudes are extracted as the perceptual features of HVS. Finally, we combined the quality indexes of both the statistical characteristics of 3D image and the perceptual properties of HVS to yield 3D image quality index. Experiments are performed on published 3D image quality assessment database show that our proposed R3DQAE model achieves highly competitive performance as compared with the state-of-the art some typical full-reference (FR) and RR models.