A Novel Image Quality Assessment Method Based on Vision Attention

This paper presents a novel variance weighted visual information fidelity (VVIF) method for full reference image quality assessment (IQA) based on the visual attention. IQA aims to use computational models to measure the image quality consistently with subjective evaluations. In psychophysiology, variance is a mechanism that can stand for the structure information of an image and then express visual attention. Thus, it directly reflects how much attention the human visual system (HVS) will distributes to the region. The distinguished visual information fidelity (VIF) method explored the relationship between image information and visual quality. In our algorithm, the variance is applied to simulate the visual attention for human beings perceiving the image sensitivity for each image block in an image. The experimental results show that the proposed algorithm has a greater consistency with the subjective assessmen.