Reduced-reference image quality assessment with orientation selectivity based visual pattern

Reduced-reference (RR) image quality assessment (IQA) method aims to accurately measure quality with part of the reference data. The challenge for RR IQA is how to effectively represent the visual content of an image with limited data for quality measurement. Inspired by the orientation selectivity (OS) mechanism in the primary visual cortex, we introduce an OS based visual pattern (OSVP) to extract visual content for RR IQA. The OS arises from the arrangement of the excitatory and inhibitory interactions among connected cortical neurons. Inspired by this, we investigate the correlation among neighbor pixels, and propose the OSVP to represent the visual content of an image. Then, the quality degradation is measured as the changes of OSVP, and a novel RR IQA model is proposed. Experimental results demonstrate that the OSVP based RR IQA model uses limited reference data (9 values) and performs highly consistent with the subjective perception.