1. Entropy coding Present by 陳群元

2. outline

3. constraints  Compression efficiency  Computational efficiency  Error robustness

4. Encoding  DCT  Reordering  Run-level coding  Entropy coding

5. Encoding  DCT  Reordering  Run-level coding  Entropy coding

6. DCT

7. Reordering  The optimum method of reordering the quantised data depends on the distribution of the non-zero coefficients.

8. Evenly distribution

9. Zigzag reordering pattern

10. Interlaced video-vary more in vertical

11. Modified reordering pattern

12. Encoding  DCT  Reordering  Run-level coding  Entropy coding

13. Run-level coding  Long sequences of identical values (zeros in this case) can be represented as a (run, level) code  (run) indicates the number of zeros preceding a non-zero value  (level) indicates the sign and magnitude of the non-zero coefficient

14. Run-level coding(ex)

15. Encoding  DCT  Reordering  Run-level coding  Entropy coding

16. Huffman Coding  ‘True’ Huffman Coding  Modified Huffman Coding  Table Design  Entropy Coding Example  Variable Length Encoder Design  Variable Length Decoder Design  Dealing with Errors

17. Huffman Coding  ‘True’ Huffman Coding  Modified Huffman Coding  Table Design  Entropy Coding Example  Variable Length Encoder Design  Variable Length Decoder Design  Dealing with Errors

18. True Huffman Coding

19. Generating the HufSman code tree  1. Order the list of data in increasing order of probability.  2. Combine the two lowest-probability data items into a ‘node’ and assign the joint probability of the data items to this node.  3. Reorder the remaining data items and node(s) in increasing order of probability and repeat step 2.

21. encoding

22. decoding

23. Carphone vs claire

24. claire

25. disadvantage  the decoder must use the same codeword set as the encoder  reduces compression efficiency.  calculating the probability table for a large video sequence a significant computational overhead  cannot be done until after the video data is encoded

26. Huffman Coding  ‘True’ Huffman Coding  Modified Huffman Coding  Table Design  Entropy Coding Example  Variable Length Encoder Design  Variable Length Decoder Design  Dealing with Errors

27. Table design Final pair

28. Table design

29. H.263/MPEG-4 motion vector difference (MVD)  The H.263MPEG-4 differentially coded motion vectors (MVD)  are each encoded as a pair of VLCs, one for the x- component and one for the y-component

30. mvd

32. H.26L universal VLC (UVLC)  The emerging H.26L standard takes a step away from individually calculated Huffman tables by using a ‘universal’ set of VLCs for any coded element. Each codeword is generated from the following systematic list:

35. Huffman Coding  ‘True’ Huffman Coding  Modified Huffman Coding  Table Design  Entropy Coding Example  Variable Length Encoder Design  Variable Length Decoder Design  Dealing with Errors

36. Entropy Coding Example

37. Huffman Coding  ‘True’ Huffman Coding  Modified Huffman Coding  Table Design  Entropy Coding Example  Variable Length Encoder Design  Variable Length Decoder Design  Dealing with Errors

38. Variable Length Encoder Design

39. Huffman Coding  ‘True’ Huffman Coding  Modified Huffman Coding  Table Design  Entropy Coding Example  Variable Length Encoder Design  Variable Length Decoder Design  Dealing with Errors

40. Variable Length Decoder Design

41. Huffman Coding  ‘True’ Huffman Coding  Modified Huffman Coding  Table Design  Entropy Coding Example  Variable Length Encoder Design  Variable Length Decoder Design  Dealing with Errors

42. Dealing with Errors

43. Arithmetic Coding

44. Ex. encode (0,-1,0,2) 0.394

46. Ex. decode (0,-1,0,2)

47. Arithmetic coding vs huffman  Ideal  0.394 in this case, which can be represented as a fixed- point number with sufficient accuracy using 9 bits  Huffman: 1;0011;1;0000110=>13 bits

48.  The end  Thank you