Encoding mode selection by optimising an estimate of distortion, based on a trained parameter, and a measure of required bit rate

摘要

A method of encoding an outgoing video stream comprising plural frames, each frame having plural image portions, comprises: for each target image portion, selecting 49 a preferred one of a set of encoding modes (e.g., intra 41, inter 43 or skip modes) by optimizing a function comprising a distortion estimate and measure of bit rate required to encode the target image portion; encoding the target image portion into the video stream using the selected encoding mode; and transmitting the encoded stream over a lossy channel. The distortion estimate comprises a first, source coding distortion term (e.g., sum of absolute differences (SAD)), and a bias term representing an estimate of distortion that would be experienced due to channel loss; the bias term is determined based on a trained parameter (e.g., a function of signal to noise ratio (SNR), loss probability, coding rate or round trip time) trained based on a sample video stream. The bias term may estimate distortion due to the non arrival of a reference portion in the target portionâ s history on which the target portion prediction depends, and on a concealment term representing distortion experienced due to concealment; the sum of squared differences (SSD) between samples may be used.