Skip to Main Content
This paper proposes a decoder side estimation for a just noticeable distortion perceptual model integrated in the H.264/AVC codec to improve its RD performance. The proposed estimation avoids having to send the perceptual weights used during the quantization of each frequency coefficient. Experiments with high definition videos show that the novel perceptual video codec is able to reduce the bitrate up to about 34% at the cost of, at most, 1.2% objective multi scale structural similarity quality loss. Moreover, the RD performance with the estimation is comparable to the performance achieved if the original perceptual model is used at the decoder without considering the associated rate.