A Reduced Reference Video Quality Assessment method based on the human motion perception

Reduced Reference Video Quality Assessment (RRVQA) methods estimate the amount of distortion in the distorted video using particular information from the original video. In this paper we propose a RRVQA based on the importance of the human perception of motion in the video. As we know from, objects' motion with different speeds has different attractions for human. We have developed a weighting map for each frame based on the motion speed of each part of a frame and combined it with a reduced reference divisive normalization Transform (DN) based image quality assessment method in order to achieve a RRVQA algorithm. In order to study the performance of this algorithm, we have used VQEG Phase I dataset and its respective MOS scores. By comparing the estimated score from our method and the MOS score, we have shown the performance of our proposed method. In addition to it, in order to further study the performance of our method, we compared our proposed method to some full reference VQA algorithms. Results show that the proposed method has an acceptable performance even between full reference methods.