摘要 |
A method, system, and computer program for coding video frames in a video compression system. One aspect includes automatically scaling coding mode biases as a function of the number of bits of coding precision in the video compression system, or dynamic range and/or contrast range, or quantization parameter. Another aspect includes selecting coding mode biases as a function of the total number of bits required for macroblock coding, or setting all biases to zero. Another aspect includes selecting one of a plurality of coding modes (1) resulting in no more than a selected number of coded bits, or (2) having at least a selected image quality, or (3) having a combination of number of coded bits and image quality. Another aspect includes selecting, as a preferred coding mode, one of a plurality of coding modes having a selected combination of number of coded bits and image quality based on various AC and DC motion vectors. Another aspect includes selecting, as a preferred frame coding choice, one of a plurality of frame coding choices having a selected combination of number of coded bits and image quality. The number of coded bits can also be determined for a set of selected quantization parameter (QP) values or quantization frequency weighting matrices, for each frame coding choice. This approach may also be used to select a preferred QP value or quantization frequency weighting matrix.
|