Video coding based on first order prediction and pre-defined second order prediction mode

摘要

This disclosure describes video encoding and decoding techniques in which a first order prediction process and a second order prediction process are used in combination to generate predictive video blocks for video coding. First order prediction may be similar to conventional motion estimation and motion compensation that generates residual video blocks. The second order prediction may involve a process similar to conventional intra-prediction, but is performed on the residual video blocks. The techniques of this disclosure may pre-define the second order prediction to a specific mode, such as a mode similar to the intra-DC mode used in intra coding. In addition, the techniques of this disclosure may combine aspects of the first order and second order prediction into a single process so that the effects of second order prediction on the residuals are taken into account during the first order prediction process, which may improve compression.

主权项

1. A method of encoding video data in a video encoder, the method comprising:
performing a first order prediction process via the video encoder to generate at least two first order residual blocks for a given video block being coded; performing a second order prediction process via the video encoder to generate at least one second order residual block for the given video block being coded based on one of the first order residual blocks and a pre-defined mode for the second order prediction process; selecting one of the second order residual block or at least one of the first order residual blocks for encoding the given video block based on a rate-distortion analysis via the video encoder with respect to the second order residual block and the at least one of the first order residual blocks to substantially minimize a rate-distortion cost associated with encoding the given video block, wherein the rate-distortion analysis with respect to the second order residual block accounts for a combined effect of the first order prediction process and the second order prediction process in coding the given video block; and encoding the given video block for transmission based on the selected one of the second order residual block or at least one of the first order residual blocks.