High precision encoding and decoding of video images

• US 20020154693A1
• Filed: 03/02/2001
• Published: 10/24/2002
• Est. Priority Date: 03/02/2001
• Status: Active Grant

• Current Assignee: Dolby Laboratories Incorporated, Dolby Laboratories Licensing Corporation (Dolby Laboratories Incorporated)
• Original Assignee: Dolby Laboratories Licensing Corporation (Dolby Laboratories Incorporated)
• Inventors: Demos, Gary A., Ruhoff, David
• Granted Patent: US 8,374,237 B2
• Time in Patent Office: Days
• US Class Current: 375/240.03
• CPC Class Codes: H04N 19/124 Quantisation
• CPC Class Codes: H04N 19/136 Incoming video signal characteristics or properties
• CPC Class Codes: H04N 19/182 the unit being a pixel
• CPC Class Codes: H04N 19/184 the unit being bits, e.g. of the compressed video stream
• CPC Class Codes: H04N 19/186 the unit being a colour or a chrominance component
• CPC Class Codes: H04N 19/196 being specially adapted for the computation of encoding parameters, e.g. by averaging previously computed encoding parameters processing of motion vectors H04N19/513
• CPC Class Codes: H04N 19/45 performing compensation of the inverse transform mismatch, e.g. Inverse Discrete Cosine Transform [IDCT] mismatch
• CPC Class Codes: H04N 19/60 using transform coding
• CPC Class Codes: H04N 19/61 in combination with predictive coding
• CPC Class Codes: H04N 19/90 using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals

Abstract

Methods, systems, and computer programs for improved quality video compression. Image quality from MPEG-style video coding may be improved by preserving a higher number of bits during intermediate encoding and decoding processing steps. Problems of inverse discrete cosine transform (IDCT) mismatch can be eliminated by exactly matching the IDCT function numerical algorithm of the decoder to the IDCT function numerical algorithm used for the decoding portion of the encoder. Also included is an application of high precision compression to wide dynamic range images by extending the range of the “quantization parameter” or “QP”. The extension of QP may be accomplished either by increasing the range of QP directly, or indirectly through a non-linear transformation. Also included is an application of extended intermediate processing precision and an extended QP range to reduced contrast regions of an image to extend the precision with which the low-contrast portions are compression coded.

Claims

1. A method for compressing a sequence of digitized video images including a sequence of frames represented at a first precision in a first color space, the method including:

(a) transforming the sequence of frames to a representation in a second color space at a second precision greater than the first precision; and

(b) performing subsequent encoding steps at the second precision to create a compressed output.

2. A method for compressing a sequence of digitized video images including a sequence of frames represented at a first precision in a first color space, the method including:

(a) transforming the sequence of frames to a representation in a second color space at a second precision greater than the first precision;

(b) performing a motion compensated discrete cosine transform at the second precision on the sequence of frames to produce a first encoded output;

(c) quantizing the first encoded output at the second precision to create a quantized output; and

(d) performing an inverse discrete cosine transform at the second precision on the quantized output to produce a compressed output at the second precision.

3. A method for compressing and decompressing a sequence of digitized video images including a sequence of frames represented at a first precision, the method including encoding the sequence of frames at a second precision greater than the first precision to create a compressed bitstream.

4. The method of claim 3, further including decoding the compressed bitstream at the second precision.

5. The method of claim 3, wherein encoding the sequence of frames includes transforming the sequence of frames to a representation in a second color space at the second precision.

6. A method for compressing and decompressing a sequence of digitized video images including a sequence of frames represented at a first precision in a first color space, the method including:

(a) transforming the sequence of frames to a representation in a second color space at a second precision greater than the first precision;

(b) performing subsequent encoding steps at the second precision to create a compressed bitstream; and

(c) decoding the compressed bitstream by;

(1) dequantizing the compressed bitstream at the second precision to create a dequantized output;

(2) applying an inverse discrete cosine transform at the second precision on the dequantized output to produce a decompressed output;

(3) generating image frames at the second precision from the decompressed output.

7. The method of claim 6, wherein the subsequent encoding steps include applying an inverse discrete cosine transform numerical algorithm, and wherein decoding the compressed bitstream includes applying a matching inverse discrete cosine transform numerical algorithm on the dequantized output.

8. The method of claim 6, further including generating P frames at the second precision.

9. The method of claim 6, further including generating B frames at the second precision.

10. A method for compressing and decompressing a sequence of digitized video images including a sequence of frames represented at a first precision in a first color space, the method including:

(a) transforming the sequence of frames to a representation in a second color space at a second precision greater than the first precision;

(b) performing subsequent encoding steps at the second precision to create a compressed bitstream, including applying an inverse discrete cosine transform numerical algorithm; and

(c) performing subsequent decoding steps at the second precision on the compressed bitstream, including applying a matching inverse discrete cosine transform numerical algorithm.

11. A method for increasing the average compression ratio during compression of a sequence of digitized video images, the digitized video images including a sequence of frames, into a compressed bitstream that includes I frames, the method including encoding the sequence of frames by applying an inverse discrete cosine transform numerical algorithm that matches the inverse discrete cosine transform numerical algorithm applied during a subsequent decoding process, thereby generating a compressed bitstream having increased spacing between I frames.

12. The method of claim 11, further including decoding the compressed bitstream, including applying an inverse discrete cosine transform numerical algorithm that matches the inverse discrete cosine transform numerical algorithm applied during encoding.

14. The method of claim 13, further including directly extending the range of possible QP values by increasing the number of bits representing QP codes.

15. The method of claim 13, further including effectively extending the range of possible QP values by mapping QP codes to a wider range of possible QP values.

16. The method of claim 13, wherein determining a QP value for each macroblock of each frame is based upon regional information with respect to one or more frames.

17. The method of claim 16, wherein the regional information includes local image region contrast within each frame.

18. The method of claim 16, wherein the regional information includes local dynamic range within each frame.

19. The method of claim 16, wherein the regional information includes local detail amplitudes within each frame.

20. The method of claim 16, wherein the regional information includes local motion between one or more sequential frames.

21. The method of claim 13, wherein determining a QP value for each macroblock of each frame is based upon information generated during the compressing step.

22. The method of claim 21, wherein the information generated during the compressing step includes relative amplitudes of discrete cosine transform coefficients for each macroblock.

23. The method of claim 21, wherein the information generated during the compressing step includes a constant number of bits assigned to encode each macroblock within a region of a frame.

24. The method of claim 13, further including:

(a) for each compressed frame, redetermining a QP value for each QP code associated with such compressed frame; and

(b) decompressing each compressed frame at the second precision to create a decompressed frame having at least one of a wide dynamic range or wide contrast range, such decompressing including applying the re-determined QP values for such frame.

28. The computer program of claim 27, further including instructions for causing the computer to decode the compressed bitstream at the second precision.

29. The computer program of claim 27, wherein the instructions for causing the computer to encode the sequence of frames includes instructions for causing the computer to transform the sequence of frames to a representation in a second color space at the second precision.

13. A method for compressing a sequence of digitized video images having at least one of a wide dynamic range or wide contrast range, the video images including frames represented at a first precision, each frame including a plurality of macroblocks defining regions, the method including:

(a) determining a quantization parameter, QP, having codes that represent either a direct correspondence to possible values or an extended range of possible values;

(b) determining a QP value for each macroblock of each frame;

(c) compressing each frame at a second precision higher than the first precision to create a compressed frame, such compressing including applying the determined QP values for such frame to reduce the number of bits required to encode such frame;

(d) associating QP codes with the determined QP values used during compressing; and

(e) outputting each compressed frame and the associated QP codes.

25. A computer program, stored on a computer-readable medium, for compressing a sequence of digitized video images including a sequence of frames represented at a first precision in a first color space, the computer program comprising instructions for causing a computer to:

(a) transform the sequence of frames to a representation in a second color space at a second precision greater than the first precision; and

(b) perform subsequent encoding steps at the second precision to create a compressed output.

26. A computer program, stored on a computer-readable medium, for compressing a sequence of digitized video images including a sequence of frames represented at a first precision in a first color space, the computer program comprising instructions for causing a computer to:

(a) transform the sequence of frames to a representation in a second color space at a second precision greater than the first precision;

(b) perform a motion compensated discrete cosine transform at the second precision on the sequence of frames to produce a first encoded output;

(c) quantize the first encoded output at the second precision to create a quantized output; and

(d) perform an inverse discrete cosine transform at the second precision on the quantized output to produce a compressed output at the second precision.

27. A computer program, stored on a computer-readable medium, for compressing and decompressing a sequence of digitized video images including a sequence of frames represented at a first precision in a first color space, the computer program comprising instructions for causing a computer to encode the sequence of frames at a second precision greater than the first precision to create a compressed bitstream.

30. A computer program, stored on a computer-readable medium, for compressing and decompressing a sequence of digitized video images including a sequence of frames represented at a first precision in a first color space, the computer program comprising instructions for causing a computer to:

(a) transform the sequence of frames to a representation in a second color space at a second precision greater than the first precision;

(b) perform subsequent encoding steps at the second precision to create a compressed bitstream; and

(c) decode the compressed bitstream by;

(1) dequantizing the compressed bitstream at the second precision to create a dequantized output;

(2) applying an inverse discrete cosine transform at the second precision on the dequantized output to produce a decompressed output;

(3) generating image frames at the second precision from the decompressed output.

31. The computer program of claim 30, wherein the subsequent encoding steps include instructions for causing the computer to apply an inverse discrete cosine transform numerical algorithm, and wherein the instructions for causing the computer to decode the compressed bitstream include instructions for causing the computer to apply a matching inverse discrete cosine transform numerical algorithm on the dequantized output.

32. The computer program of claim 31, further including instructions for causing the computer to generate P frames at the second precision.

33. The computer program of claim 31, further including instructions for causing the computer to generate B frames at the second precision.

36. The computer program of claim 35, further including instructions for causing the computer to decode the compressed bitstream, including instructions for causing the computer to apply an inverse discrete cosine transform numerical algorithm that matches the inverse discrete cosine transform numerical algorithm applied during encoding.

38. The computer program of claim 37, further including instructions for causing the computer to directly extend the range of possible QP values by increasing the number of bits representing QP codes.

39. The computer program of claim 37, further including instructions for causing the computer to effectively extend the range of possible QP values by mapping QP codes to a wider range of possible QP values.

40. The computer program of claim 37, wherein determining a QP value for each macroblock of each frame is based upon regional information with respect to one or more frames.

41. The computer program of claim 40, wherein the regional information includes local image region contrast within each frame.

42. The computer program of claim 40, wherein the regional information includes local dynamic range within each frame.

43. The computer program of claim 40, wherein the regional information includes local detail amplitudes within each frame.

44. The computer program of claim 40, wherein the regional information includes local motion between one or more sequential frames.

45. The computer program of claim 37, wherein determining a QP value for each macroblock of each frame is based upon information generated during the compressing step.

46. The computer program of claim 45, wherein the information generated during the compressing step includes relative amplitudes of discrete cosine transform coefficients for each macroblock.

47. The computer program of claim 45, wherein the information generated during the compressing step includes a constant number of bits assigned to encode each macroblock within a region of a frame.

48. The computer program of claim 47, further including instructions for causing the computer to:

(a) for each compressed frame, re-determine a QP value for each QP code associated with such compressed frame; and

(b) decompress each compressed frame at the second precision to create a decompressed frame having at least one of a wide dynamic range or wide contrast range, such decompression including applying the re-determined QP values for such frame.

34. A computer program, stored on a computer-readable medium, for compressing and decompressing a sequence of digitized video images including a sequence of frames represented at a first precision in a first color space, the computer program comprising instructions for causing a computer to:

(a) transform the sequence of frames to a representation in a second color space at a second precision greater than the first precision;

(b) perform subsequent encoding steps at the second precision to create a compressed bitstream, including applying an inverse discrete cosine transform numerical algorithm; and

(c) perform subsequent decoding steps at the second precision on the compressed bitstream, including applying a matching inverse discrete cosine transform numerical algorithm.

35. A computer program, stored on a computer-readable medium, for increasing the average compression ratio during compression of a sequence of digitized video images, the digitized video images including a sequence of frames, into a compressed bitstream that includes I frames, the computer program comprising instructions for causing a computer to encode the sequence of frames by applying an inverse discrete cosine transform numerical algorithm that matches the inverse discrete cosine transform numerical algorithm applied during a subsequent decoding process, thereby generating a compressed bitstream having increased spacing between I frames.

37. A computer program, stored on a computer-readable medium, for compressing a sequence of digitized video images having at least one of a wide dynamic range or wide contrast range, the video images including frames represented at a first precision, each frame including a plurality of macroblocks defining regions, the computer program comprising instructions for causing a computer to:

(a) determine a quantization parameter, QP, having codes that represent either a direct correspondence to possible values or an extended range of possible values;

(b) determine a QP value for each macroblock of each frame;

(c) compress each frame at a second precision higher than the first precision to create a compressed frame, such compressing including applying the determined QP values for such frame to reduce the number of bits required to encode such frame;

(d) associate QP codes with the determined QP values used during compressing; and

(e) output each compressed frame and the associated QP codes.

49. A system for compressing a sequence of digitized video images including a sequence of frames represented at a first precision in a first color space, the system including:

(a) means for transforming the sequence of frames to a representation in a second color space at a second precision greater than the first precision; and

(b) means for performing subsequent encoding steps at the second precision to create a compressed output.

50. A system for compressing a sequence of digitized video images including a sequence of frames represented at a first precision in a first color space, the system including:

(a) means for transforming the sequence of frames to a representation in a second color space at a second precision greater than the first precision;

(b) means for performing a motion compensated discrete cosine transform at the second precision on the sequence of frames to produce a first encoded output;

(c) means for quantizing the first encoded output at the second precision to create a quantized output; and

(d) means for performing an inverse discrete cosine transform at the second precision on the quantized output to produce a compressed output at the second precision.

51. A system for compressing and decompressing a sequence of digitized video images including a sequence of frames represented at a first precision in a first color space, the system including means for encoding the sequence of frames at a second precision greater than the first precision to create a compressed bitstream.

52. The system of claim 51, further including means for decoding the compressed bitstream at the second precision.

53. The system of claim 51, wherein the means for encoding the sequence of frames includes means for transforming the sequence of frames to a representation in a second color space at the second precision.

54. A system for compressing and decompressing a sequence of digitized video images including a sequence of frames represented at a first precision in a first color space, the system including:

(a) means for transforming the sequence of frames to a representation in a second color space at a second precision greater than the first precision;

(b) means for performing subsequent encoding steps at the second precision to create a compressed bitstream; and

(c) means for decoding the compressed bitstream by;

(1) dequantizing the compressed bitstream at the second precision to create a dequantized output;

(2) applying an inverse discrete cosine transform at the second precision on the dequantized output to produce a decompressed output;

(3) generating image frames at the second precision from the decompressed output.

55. The system of claim 54, wherein the means for performing subsequent encoding steps include means for applying an inverse discrete cosine transform numerical algorithm, and wherein the means for decoding the compressed bitstream includes means for applying a matching inverse discrete cosine transform numerical algorithm on the dequantized output.

56. The system of claim 54, further including means for generating P frames at the second precision.

57. The system of claim 54, further including means for generating B frames at the second precision.

60. The system of claim 59, further including means for decoding the compressed bitstream, including means for applying an inverse discrete cosine transform numerical algorithm that matches the inverse discrete cosine transform numerical algorithm applied during encoding.

58. A system for compressing and decompressing a sequence of digitized video images including a sequence of frames represented at a first precision in a first color space, the system including:

(a) means for transforming the sequence of frames to a representation in a second color space at a second precision greater than the first precision;

(b) means for performing subsequent encoding steps at the second precision to create a compressed bitstream, including applying an inverse discrete cosine transform numerical algorithm; and

(c) means for performing subsequent decoding steps at the second precision on the compressed bitstream, including applying a matching inverse discrete cosine transform numerical algorithm.

59. A system for increasing the average compression ratio during compression of a sequence of digitized video images, the digitized video images including a sequence of frames, into a compressed bitstream that includes I frames, the system including means for encoding the sequence of frames by applying an inverse discrete cosine transform numerical algorithm that matches the inverse discrete cosine transform numerical algorithm applied during a subsequent decoding process, thereby generating a compressed bitstream having increased spacing between I frames.

61. A system for compressing a sequence of digitized video images having at least one of a wide dynamic range or wide contrast range, the video images including frames represented at a first precision, each frame including a plurality of macroblocks defining regions, the system including:

(a) means for determining a quantization parameter, QP, having codes that represent either a direct correspondence to possible values or an extended range of possible values;

(b) means for determining a QP value for each macroblock of each frame;

(c) means for compressing each frame at a second precision higher than the first precision to create a compressed frame, including means for applying the determined QP values for such frame to reduce the number of bits required to encode such frame;

(d) means for associating QP codes with the determined QP values used during compressing; and

(e) means for outputting each compressed frame and the associated QP codes.

62. The system of claim 61, further including means for directly extending the range of possible QP values by increasing the number of bits representing QP codes.

63. The system of claim 61, further including means for effectively extending the range of possible QP values by mapping QP codes to a wider range of possible QP values.

64. The system of claim 61, wherein the means for determining a QP value for each macroblock of each frame includes means for making such a determination based upon regional information with respect to one or more frames.

65. The system of claim 64, wherein the regional information includes local image region contrast within each frame.

66. The system of claim 64, wherein the regional information includes local dynamic range within each frame.

67. The system of claim 64, wherein the regional information includes local detail amplitudes within each frame.

68. The system of claim 64, wherein the regional information includes local motion between one or more sequential frames.

69. The system of claim 61, wherein the means for determining a QP value for each macroblock of each frame includes means for making such a determination based upon information generated during the compressing step.

70. The system of claim 69, wherein the information generated by the means for compressing includes relative amplitudes of discrete cosine transform coefficients for each macroblock.

71. The system of claim 69, wherein the information generated by the means for compressing includes a constant number of bits assigned to encode each macroblock within a region of a frame.

72. The system of claim 61, further including:

(a) means for re-determining a QP value for each QP code associated with each compressed frame; and

(b) means for decompressing each compressed frame at the second precision to create a decompressed frame having at least one of a wide dynamic range or wide contrast range, such decompressing including applying the re-determined QP values for such frame.