| 摘要 |
In many videoconferencing applications, bandwidth is at a premium, and thus, it is important to encode a given video frame intelligently. It is often desirable that a larger amount of information be spent encoding the more important parts of the video frame, e.g., human facial features, whereas the less important parts of the video frame can be compressed at higher rates. Thus, there is need for an apparatus, computer readable medium, processor, and method for intelligent skin tone and facial feature aware videoconferencing compression that can "suggest" intelligent macroblock compression ratios to a video encoder. The suggestion of compression rates can be based at least in part on a determination of which macroblocks in a given video frame are likely to contain skin tones, likely to contain features (e.g., edges), likely to contain features in or near skin tone regions, or likely to contain neither skin tones nor features. |
