Reduced-Reference Visual Quality Assessment Using Central Moments of Edge Projections

Objective visual quality assessment algorithms quantify the visual difference between a reference image and the distorted version as human being perceives. The visual quality assessment method using central moments of edge projections is presented as a feature-based reduced-reference approach. Edges of the reference video are different from those of the distorted one due to blur, blocky, or other artifacts. The edge map of each frame in a video is separately projected along horizontal/vertical directions that are represented in terms of a number of central moments. In experiments, 20 central moments are used as features for evaluating the visual quality of distorted videos. The effectiveness of the proposed algorithm is shown with abundant sets of video sequences.