Video entropy decoding with graceful degradation

摘要

A system and method are provided for gracefully degrading video quality in a video entropy decoding system. The method comprises: accepting a sequence of coded video frames; measuring the speed at which the video frames are decoded; in response to the decoding speed, varying the degree of data decoded in each macroblock (MB) of a frame; and, supplying decoded video data from every MB in the frame. Though the degree of data decoded in each MB of the frame may vary, predictions from a first MB in a frame can still be used to decode a second MB in the frame. That is, synchronization is maintained between the first and second MB. Varying the degree of data decoded in each MB of the frame includes: decoding a minimal amount of data in each MB in the frame; and, varying the level of decoding performed on each MB in the frame.

主权项

1. A method for gracefully degrading video quality in a video entropy decoding system, the method comprising:
a decoder accepting a sequence of coded video frames; a resource monitor measuring decoding processing power required to decode the video frames; in response to the required decoding processing power, the decoder varying the level of decoding processing for each macroblock (MB) of the video frame, where the level of decoding processing is sufficient to decode at least a minimal amount of data in each MB as follows:
completely decoding each symbol in the MB using an algorithm selected from the group including Variable Length Code (VLC) and Context Adaptive VLC (CAVLC);selectively performing coefficient reconstruction; and the decoder supplying decoded video data from every MB in the frame.