Architecture for hybrid video codec

摘要

A method for video encoding comprising transforming a plurality of pixels to generate a first plurality of transform coefficients, quantizing the first plurality of transform coefficients to generate a plurality of quantized transform coefficients, computing a plurality of prediction pixels based on a plurality of reconstructed pixels which are prediction references for the plurality of pixels, transforming the plurality of prediction pixels to generate a second plurality of transform coefficients, and computing a plurality of prediction residual coefficients as a difference between the plurality of quantized transform coefficients and the second plurality of transform coefficients.

主权项

1. A method for video encoding comprising:
transforming a plurality of pixels to generate a first plurality of transform coefficients; quantizing the first plurality of transform coefficients to generate a plurality of quantized transform coefficients; computing a plurality of prediction pixels based on a plurality of reconstructed pixels which are prediction references for the plurality of pixels; transforming the plurality of prediction pixels to generate a second plurality of transform coefficients; computing a plurality of prediction residual coefficients as a difference between the plurality of quantized transform coefficients and the second plurality of transform coefficients, wherein the second plurality of transform coefficients are directly used, without any quantization, to compute the plurality of prediction residual coefficients; and entropy encoding the plurality of prediction residual coefficients to generate a plurality of encoded residual coefficients.