KC视频编码器原理 JPEG common code words e.g.,DCT: shorter symbols spatial→frequency Huffman,arithmetic coding Frames of Motion Transform, Digital Video Estimation Quantization,Zig- Symbol & Zag Scan Run- Encoder Compensation Length Encoding Bit Stream predict current Quantization changes representation size for each symbol frame from previous >adjust rate/quality trade-off Run-length encoding. long runs of zeros>run-length symbol MPEG,H.26x (ynh,cxh @ustc.edu.cn{ynh,cxh}@ustc.edu.cn 视频编码器原理 6 Motion Estimation & Compensation Transform, Quantization, Zig￾Zag Scan & Run￾Length Encoding Symbol Encoder Frames of Digital Video Bit Stream common code words  shorter symbols Huffman, arithmetic coding e.g., DCT: spatial  frequency Quantization changes representation size for each symbol adjust rate/quality trade-off Run-length encoding: long runs of zeros  run-length symbol predict current frame from previous JPEG MPEG, H.26x