ABSTRACT
Based on the research on HVS, we find a novel perspective to assess image blur: Relative comparison. Recently, we proposed two methods from the angle of relative comparison by taking HVS into account. They first construct a new image with filter as the reference image and then compute the change of a statistics constructed by ourselves between the original image and the reference image. The two methods have been testified by sets of experiment to fulfill the performance requirements and have the advantage of simplicity and efficiency. In contrast to existing popular traditional metrics, our method shows a high degree of stability for different image contents. Being advantageous in blur assessment, the new kind of methods not only can solve the problem mentioned above but also provide a novel perspective to image blur assessment study, therefore making it worth studying.
