PREDICTIVE VIDEO ENCODING

摘要

A system and method is disclosed for representing each frame in a sequence of video frames in terms of a correlation between successive video frames. The system includes a compressor (12') which generates a data representation of a current frame in current frame buffer (16) with reference to a previous frame in a previous buffer (30). The data representation is entropy encoded before being transmitted by transmitter (14) to a decompressor. The entropy encoded data representation is also provided to a decompressor (28') which decompresses the data representation to generate a frame for storage in the previous frame buffer (30). The encoded data representation sent by transmitter (14) is received by a receiver (18), decoded by decoder (99) and decompressed by decompressor (20') using a previous decompressed frame stored in previous frame buffer (24) to generate a current frame stored in current frame buffer (80). Decompressor (25) may also include components for filtering the current frame to remove artifacts in the image and the filtered frame is stored in filtered current frame buffer (82). The filtered frame stored in filtered current frame buffer (82) is also provided to previous frame buffer (24) so the decompressor (20') uses the same previous frame data used by compressor (12'). If decompressor (28') includes the filtering components, a filtered frame is also provided to previous frame buffer (30). The compression method implemented by compressor (12') is a novel motion vector method in which the data representation of a current frame includes a shape value, coefficient values, a differential motion vector, a range area adjustment value, and a block size value.