Perceptully lossless video re-encoding for cloud transcoding

Abstract

In this paper, we present a perceptually lossless video re-encoding approach to cloud transcoding based on a just noticeable distortion (JND) model. The bitrate is minimized by adaptive step size and dynamic rounding offset adjustment. On the average, the proposed approach achieves 9.7% overall bitrate reduction compared to the H.264/AVC JM reference encoder under the same coding condition. The performance of the resulting cloud transcoding is further verified by a subjective test.