Binocular combination and fractional differential based 3D image quality assessment

摘要

Three dimensional image quality assessment (3DIQA) is an important issue for three-dimensional video technologies and faces more challenges than its two-dimensional counterparts. In this paper, a novel 3DIQA method is proposed based on human visual system considering the influence of binocular combination between the left and right views of stereo image. To be more specific, difference-of-Gaussian response is defined and used to construct cyclopean images of stereo images. Also, fractional differential masks are defined instead of the gradient operator to improve image local feature. Then, a fractional differential-based feature similarity metric is used to derive a single quality score. Experimental results demonstrate that the proposed method achieves better consistency with subjective quality assessment compared with other representative methods.