Hierarchical encoding and decoding apparatus for a digital image signal

• US 5,663,764 A
• Filed: 07/19/1995
• Issued: 09/02/1997
• Est. Priority Date: 09/30/1993
• Status: Expired due to Term

• Current Assignee: Sony Corporation (Sony Group Corp.)
• Original Assignee: Sony Corporation (Sony Group Corp.)
• Inventors: Kondo, Tetsujiro, Fujimori, Yasuhiro, Nishikata, Takeharu
• Primary Examiner(s): Britton, Howard W.
• Application Number:
  US08/504,040
• Time in Patent Office: 776 Days
• Field of Search: 348/409, 348/410, 348/411, 348/412, 348/414, 348/417, 348/418, 348/421, 348/424
• US Class Current: 375/240.14
• CPC Class Codes: H04N 19/102 characterised by the element, parameter or selection affected or controlled by the adaptive coding
• CPC Class Codes: H04N 19/61 in combination with predictive coding
• CPC Class Codes: H04N 19/63 using sub-band based transform, e.g. wavelets
• CPC Class Codes: H04N 19/645 by grouping of coefficients into blocks after the transform
• CPC Class Codes: H04N 19/94 Vector quantisation

Abstract

In a hierarchical encoding apparatus, predictors each including a look-up table are used to form predicted signals for respective hierarchical levels. The difference between the predicted signal and the original signal forms the differential signal for the higher hierarchical levels. The look-up table contains either predictive coefficients which are used as weights for pixels of a signal in one hierarchical level to produce a signal at a higher hierarchical level, or representative values which are combined with a dynamic range and redundant component in a set of pixels of a signal in one hierarchical level to produce a signal at a higher hierarchical level. The values in the look-up table are obtained from representative images.

Claims

1. A hierarchical encoding apparatus for encoding a first digital image signal representing pixels and having a first resolution, comprising:

means for forming a second digital image signal from the first digital image signal, the second digital image signal having a second resolution lower than the first resolution;

predicting means for producing predicted values of the first digital image signal from the second digital image signal;

means for subtracting a predicted value of the first digital image signal from the pixel value of the first digital image signal to produce a differential value; and

means for transmitting the second digital image signal and a differential signal representing the differential values;

wherein the predicting means comprises;

class categorizing means for determining a class corresponding to a selected pixel of the second digital image signal;

the class categorizing means being operative to produce a plurality of predicted pixel values of the selected pixel, each of the predicted pixel values corresponding to a respective one of a plurality of predetermined classes;

the class categorizing means being further operative to select a class corresponding to the selected pixel from the plurality of predetermined classes based on a respective one of the plurality of predicted pixel values which is closest to a value of the selected pixel; and

predicted value generating means for generating a predicted value of the first digital image signal based on the selected class.

2. The apparatus of claim 1, wherein the predicted value generating means comprises:

means for producing the predicted value based on at least one data predicting coefficient and a value of at least one pixel of the second digital image signal; and

supplying means for supplying the at least one data predicting coefficient based on the selected class.

3. The apparatus of claim 2, wherein the supplying means comprises a memory storing the at least one data predicting coefficient, the at least one data predicting coefficient being produced based on a minimum difference between a value of a prelearned pixel of a standard digital image signal having a resolution corresponding to the second resolution and a linear combination of pixel values of pixels of a standard digital image signal having a resolution corresponding to the second resolution spatially and/or temporally adjacent to the prelearned pixel.

4. The apparatus of claim 1, wherein the class categorizing means comprises:

linear combining means for producing each of the plurality of predicted pixel values as a linear combination of the products of each of a plurality of reference pixels of the second digital image signal and a respective one of a plurality of class predicting coefficients of the respective one of the plurality of predetermined classes, the plurality of reference pixels being spatially and/or temporally adjacent to the selected pixel; and

class predicting coefficient storage means for storing the plurality of class predicting coefficients.

5. An apparatus for decoding a hierarchically encoded first digital image signal representing pixels and having a first resolution, the first digital image signal being encoded by forming a second digital image signal and a differential signal representing differential values from the first digital image signal, the second digital image signal having a second resolution lower than the first resolution, producing predicted values of the first digital image signal from the second digital image signal and subtracting the predicted values of the first digital image signal from the second digital image signal to produce the differential values, comprising:

predicting means for producing a predicted value of the first digital image signal from the second digital image signal; and

means for adding the predicted value and a corresponding one of the differential values to produce a value of the first digital image signal;

the predicting means comprising;

class categorizing means for determining a class corresponding to a selected pixel of the second digital image signal;

the class categorizing means being operative to produce a plurality of predicted pixel values of the selected pixel, each of the predicted pixel values corresponding to a respective one of a plurality of predetermined classes;

the class categorizing means being further operative to select a class corresponding to the selected pixel from the plurality of predetermined classes based on a respective one of the plurality of predicted pixel values which is closest to a value of the selected pixel; and

predicted value generating means for generating a predicted value of the first digital image signal based on the selected class.

6. The apparatus of claim 5, wherein the predicted value generating means comprises:

means for producing the predicted value based on at least one data predicting coefficient and a value of at least one pixel of the second digital image signal; and

supplying means for supplying the at least one data predicting coefficient based on the selected class.

7. The apparatus of claim 6, wherein the supplying means comprises a memory storing the at least one data predicting coefficient, the at least one data predicting coefficient being produced based on a minimum difference between a value of a prelearned pixel of a standard digital image signal having a resolution corresponding to the first resolution and a linear combination of pixel values of pixels of a standard digital image signal having a resolution corresponding to the second resolution spatially and/or temporally adjacent to the prelearned pixel.

8. The apparatus of claim 5, wherein the class categorizing means comprises:

linear combining means for producing each of the plurality of predicted pixel values as a linear combination of the products of each of a plurality of reference pixels of the second digital image signal and a respective one of a plurality of class predicting coefficients of the respective one of the plurality of predetermined classes, the plurality of reference pixels being spatially and/or temporally adjacent to the selected pixel; and

class predicting coefficient storage means for storing the plurality of class predicting coefficients.

9. A hierarchical encoding method for encoding a first digital image signal representing pixels and having a first resolution, comprising the steps of:

forming a second digital image signal from the first digital image signal, the second digital image signal having a second resolution lower than the first resolution;

producing predicted values of the first digital image signal from the second digital image signal;

subtracting a predicted value of the first digital image signal from the pixel value of the first digital image signal to produce a differential value; and

transmitting the second digital image signal and a differential signal representing the differential values;

wherein the step of producing predicted values is comprised of the steps of;

determining a class corresponding to a selected pixel of the second digital image signal;

producing a plurality of predicted pixel values of the selected pixel, each of the predicted pixel values corresponding to a respective one of a plurality of predetermined classes;

selecting a class corresponding to the selected pixel from the plurality of predetermined classes based on a respective one of the plurality of predicted pixel values which is closest to a value of the selected pixel; and

generating a predicted value of the first digital image signal based on the selected class.

10. A method of decoding a hierarchically encoded first digital image signal representing pixels and having a first resolution, the first digital image signal being encoded by forming a second digital image signal and a differential signal representing differential values from the first digital image signal, the second digital image signal having a second resolution lower than the first resolution, producing predicted values of the first digital image signal from the second digital image signal and subtracting the predicted values of the first digital image signal from the second digital image signal to produce the differential values, comprising the steps of:

producing a predicted value of the first digital image signal from the second digital image signal; and

adding the predicted value and a corresponding one of the differential values to produce a value of the first digital image signal;

the step of producing a predicted value comprising the steps of;

determining a class corresponding to a selected pixel of the second digital image signal;

producing a plurality of predicted pixel values of the selected pixel, each of the predicted pixel values corresponding to a respective one of a plurality of predetermined classes;

selecting a class corresponding to the selected pixel from the plurality of predetermined classes based on a respective one of the plurality of predicted pixel values which is closest to a value of the selected pixel; and

generating a predicted value of the first digital image signal based on the selected class.

11. A hierarchical encoding apparatus for encoding a first image signal representing pixels and having a first resolution, the apparatus comprising:

means for forming a second image signal from the first image signal, the second image signal having a second resolution lower than the first resolution;

predicting means for producing predicted pixels of the first image signal from the second image signal;

means for generating a differential value between the pixel of the first image signal and the predicted pixel of the first image signal to produce a differential signal; and

means for transmitting the second image signal and the differential signal;

wherein the predicting means comprises;

a memory;

class categorizing means for forming predicted pixels for a pixel to be predicted of the second image signal from a plurality of pixels of the second image signal and each set of class predicted coefficients of respective classes from the memory and determining a class from the predicted pixels and the pixel to be predicted as a class of a considered pixel of the first image signal, the plurality of pixels being disposed adjacent the pixel to be predicted of the second image signal;

means for generating a set of data predicted coefficients in response to the determined class, each set of data predicted coefficients for respective classes is formed by using a standard image signal having a resolution corresponding to the first resolution of the first image signal; and

predicted pixel generating means for generating a predicted pixel corresponding to the considered pixel from a plurality of pixels of the second image signal and the generated set of the data predicted coefficients.

12. A hierarchical encoding apparatus according to claim 11, wherein each class predicted coefficients for the respective classes is generated by using a standard image signal having a resolution corresponding to the second resolution of the second image signal.

13. A hierarchical encoding apparatus according to claim 12, wherein the class categorizing means forms the predicted pixels for the pixel to be predicted of the second image signal by calculating a linear combination of the plurality of pixels of the second image signal and each set of class predicted coefficients of respective classes from the memory and determines a class corresponding to a set of class predicted coefficients to form the predicted pixel which is closest to the pixel to be predicted, in the formed predicted pixels, as a class of the considered pixel.

14. A hierarchical encoding apparatus according to claim 11, wherein each set of data predicted coefficients for the respective classes is stored in a memory.

15. A hierarchical encoding apparatus according to claim 14, wherein each set of data predicted coefficients for the respective classes is generated by using a method of least squares, and the predicted pixel generating means calculates a linear combination of a plurality of pixels of the second image signal and the set of predicted coefficients from the memory to generate the considered pixel.

16. An apparatus for decoding a hierarchically encoded first image signal and producing a first image signal representing pixels having a first resolution, the first image signal being encoded by forming a second image signal and a differential signal from the first image signal, the second image signal having a second resolution lower than the first resolution, the differential signal being formed by generating a differential value between a pixel of the first image signal and a predicted pixel of the first image signal produced from the second image signal, the apparatus comprising:

means for receiving the second image signal and the differential signal;

predicting means for producing predicted pixels of the first image signal from the second image signal; and

means for adding the predicted pixel of the second image signal and a corresponding differential value of the differential signal to produce the first image;

wherein the predicting means comprises;

a memory;

class categorizing means for forming predicted pixels for a pixel to be predicted of the second image signal from a plurality of pixels of the second image signal and each set of class predicted coefficients of respective classes from the memory and determining a class from the predicted pixels and the pixel to be predicted as a class of a considered pixel of the first image signal, the plurality of pixels being disposed adjacent the pixel to be predicted of the second image signal;

means for generating a set of data predicted coefficients in response to the determined class, each set of data predicted coefficients for respective classes is formed by using a standard image signal having a resolution corresponding to the first resolution of the first image signal; and

predicted pixel generating means for generating a predicted pixel corresponding to the considered pixel from a plurality of pixels of the second image signal and the generated set of the data predicted coefficients.

17. An apparatus according to claim 16, wherein each class predicted coefficients for the respective classes is generated by using a standard image signal having a resolution corresponding to the second resolution of the second image signal.

18. An apparatus according to claim 17, wherein the class categorizing means forms the predicted pixels for the pixel to be predicted of the second image signal by calculating a linear combination of the plurality of pixels of the second image signal and each set of class predicted coefficients of the respective classes from the memory and determines a class corresponding to a set of class predicted coefficients to form the predicted pixel which is closest to the pixel to be predicted, in the formed predicted pixels, as a class of the considered pixel.

19. An apparatus according to claim 16, wherein each set of data predicted coefficients for the respective classes is stored in a memory.

20. An apparatus according to claim 16, wherein each set of data predicted coefficients for the respective classes is generated by using a method of least squares, and the predicted pixel generating means calculates a linear combination of a plurality of pixels of the second image signal and the set of predicted coefficients from the memory to generate the considered pixel.

21. A hierarchical encoding apparatus for encoding a first image signal representing pixels and having a first resolution, the apparatus comprising:

means for forming a second image signal from the first image signal, the second image signal having a second resolution lower than the first resolution;

predicting means for producing predicted pixels of the first image signal from the second image signal;

means for generating a differential value between the pixel of the first image signal and the predicted pixel of the first image signal to a differential signal; and

means for transmitting the second image signal and the differential signal;

wherein the predicting means comprises;

a memory;

class categorizing means for forming predicted pixels for a pixel to be predicted of the second image signal from a plurality of pixels of the second image signal and each set of class predicted coefficients of respective classes from the memory and determining a class from the predicted pixels and the pixel to be predicted as a class of a considered pixel of the first image signal, the plurality of pixels being disposed adjacent the pixel to be predicted of the second image signal; and

means for generating a predicted pixel corresponding to the considered pixel in response to the determined class, each predicted pixel for respective classes is formed by using a standard image signal having a resolution corresponding to the first resolution of the first image signal.

22. A hierarchical encoding apparatus according to claim 21, wherein each class predicted coefficients for the respective classes is generated by using a standard image signal having a resolution corresponding to the second resolution of the second image signal.

23. A hierarchical encoding apparatus according to claim 22, wherein the class categorizing means forms the predicted pixels for the pixel to be predicted of the second image signal by calculating a linear combination of the plurality of pixels of the second image signal and each set of class predicted coefficients of the respective classes from the memory and determines a class corresponding to a set of class predicted coefficients to form the predicted pixel which is closest to the pixel to be predicted, in the formed predicted pixels, as a class of the considered pixel.

24. A hierarchical encoding apparatus according to claim 21, wherein each predicted pixel for the respective classes is stored in a memory.

25. An apparatus for decoding a hierarchically encoded first image signal and producing a first image signal representing pixels having a first resolution, the first image being encoded by forming a second image signal and a differential signal from the first image signal, the second image signal having a second resolution lower than the first resolution, the differential signal being formed by generating a differential value between the pixel of the first image signal and a predicted pixel of the first image signal produced from the second image signal, the apparatus comprising:

means for receiving the second image signal and the differential signal;

predicting means for producing predicted pixels of the first image signal from the second image signal; and

means for adding the predicted pixel of the second image signal and a corresponding differential value of the differential signal to produce the first image;

wherein the predicting means comprises;

a memory

class categorizing means for forming predicted pixels for a pixel to be predicted of the second image signal from a plurality of pixels of the second image signal and each set of class predicted coefficients of respective classes from the memory and determining a class from the predicted pixels and the pixel to be predicted as a class of a considered pixel of the first image signal, the plurality of pixels being disposed adjacent the pixel to be predicted of the second image signal; and

means for generating a predicted pixel corresponding to the considered pixel in response to the determined class, each predicted pixel for respective classes is formed by using a standard image signal having a resolution corresponding to the first resolution of the first image signal.

26. An apparatus according to claim 25, wherein each class predicted coefficients for the respective classes is generated by using a standard image signal having a resolution corresponding to the second resolution of the second image signal.

27. An apparatus according to claim 26, wherein the class categorizing means forms the predicted pixels for the pixel to be predicted of the second image signal by calculating a linear combination of the plurality of pixels of the second image signal and each set of class predicted coefficients of the respective classes from the memory and determines a class corresponding to a set of class predicted coefficients to form the predicted pixel which is closest to the pixel to be predicted, in the formed predicted pixels, as a class of the considered pixel.

28. An apparatus according to claim 25, wherein each predicted pixel for the respective classes is stored in a memory.

29. A hierarchical encoding apparatus for encoding a first image signal representing pixels and having a first resolution, the apparatus comprising:

means for forming a second image signal from the first image signal, the second image signal having a second resolution lower than the first resolution;

predicting means for producing predicted pixels of the first image signal from the second image signal;

means for generating a differential value between the pixel of the first image signal and the predicted pixel of the first image signal to a differential signal; and

means for transmitting the second image signal and the differential signal;

wherein the predicting means comprises;

a memory;

class categorizing means for forming predicted pixels for a pixel to be predicted of the second image signal from a plurality of pixels of the second image signal and each set of class predicted coefficients of respective classes from the memory and determining a class from the predicted pixels and the pixel to be predicted as a class of a considered pixel of the first image signal, the plurality of pixels being disposed adjacent the pixel to be predicted of the second image signal; and

means for generating a predicted data corresponding to the considered pixel in response to the determined class and producing a predicated image signal corresponding to the first image signal from the predicted data, each predicted data for respective classes is formed by using a standard image signal having a resolution corresponding to the first resolution of the first image signal.

30. A hierarchical encoding apparatus according to claim 29, wherein each class predicted coefficients for the respective classes is generated by using a standard image signal having a resolution corresponding to the second resolution of the second image signal.

31. A hierarchical encoding apparatus according to claim 30, wherein the class categorizing means forms the predicted pixels for the pixel to be predicted of the second image signal by calculating a linear combination of the plurality of pixels of the second image signal and each set of class predicted coefficients of the respective classes from the memory and determines a class corresponding to a set of class predicted coefficients to form the predicted pixel which is closest to the pixel to be predicted, in the formed predicted pixels, as a class of the considered pixel.

32. A hierarchical encoding apparatus according to claim 29, wherein each predicted data for the respective classes is stored in a memory.

33. An apparatus for decoding a hierarchically encoded first image signal and producing a first image signal representing pixels and a first resolution, the first image signal being encoded by forming a second image signal and a differential signal from the first image signal, the second image signal having a second resolution lower than the first resolution, the differential signal being formed by generating a differential value between the pixel of the first image signal and predicted pixel of the first image signal produced from the second image signal, the apparatus comprising:

means for receiving the second image signal and the differential signal;

predicting means for producing predicted pixels of the first image signal from the second image signal; and

means for adding the predicted pixel of the second image signal and a corresponding differential value of the differential signal to producing the first image

wherein the predicting means comprises;

a memory;

class categorizing means for forming predicted pixels for a pixel to be predicted of the second image signal from a plurality of pixels of the second image signal and each set of class predicted coefficients of respective classes from the memory and determining a class from the predicted pixels and the pixel to be predicted as a class of a considered pixel of the first image signal, the plurality of pixels being disposed adjacent the pixel to be predicted of the second image signal;

means for generating a predicted data corresponding to the considered pixel in response to the determined class and producing a predicted image signal corresponding to the first image signal from the predicted data, each predicted data for respective classes is formed by using a standard image signal having a resolution corresponding to the first resolution of the first image signal.

34. An apparatus according to claim 33, wherein each class predicted coefficients for the respective classes is generated by using a standard image signal having a resolution corresponding to the second resolution of the second image signal.

35. An apparatus according to claim 34, wherein the class categorizing means forms the predicted pixels for the pixel to be predicted of the second image signal by calculating a linear combination of the plurality of pixels of the second image signal and each set of class predicted coefficients of the respective classes from the memory and determines a class corresponding to a set of class predicted coefficients to form the predicted pixel which is closest to the pixel to be predicted, in the formed predicted pixels, as a class of the considered pixel.

36. An apparatus according to claim 33, wherein each predicted data for the respective classes is stored in a memory.

37. A hierarchical encoding method for encoding a first image signal representing pixels and having a first resolution, the method comprising the steps of:

forming a second image signal from the first image signal, the second image signal having a second resolution lower than the first resolution;

producing predicted pixels of the first image signal from the second image signal;

generating a differential value between the pixel of the first image signal and the predicted pixel of the first image signal to produce a differential signal; and

transmitting the second image signal and the differential signal;

wherein the predicting step comprises the step of;

forming predicted pixels for a pixel to be predicted of the second image signal from a plurality of pixels of the second image signal and each set of class predicted coefficients of respective classes from the memory, the plurality of pixels being disposed adjacent the pixel to be predicted of the second image signal;

determining a class from the predicted pixels and the pixel to be predicted as a class of a considered pixel of the first image signal;

generating a set of data predicted coefficients in response to the determined class, each set of data predicted coefficients for respective classes is formed by using a standard image signal having a resolution corresponding to the first resolution of the first image signal; and

generating a predicted pixel corresponding to the considered pixel from a plurality of pixels of the second image signal and the generated set of the data predicted coefficients.

38. A hierarchical encoding method according to claim 37, wherein each class predicted coefficients for the respective classes is generated by using a standard image signal having a resolution corresponding to the second resolution of the second image signal.

39. A hierarchical encoding method according to claim 38, wherein the predicted pixels forming step forms the predicted pixels for the pixel to be predicted of the second image signal by calculating a linear combination of the plurality of pixels of the second image signal and each set of class predicted coefficients of respective classes from the memory;

and the class determining step determines a class corresponding to a set of class predicted coefficients to form the predicted pixel which is closest to the pixel to be predicted, in the formed predicted pixels, as a class of the considered pixel.

40. A hierarchical encoding method according to claim 37, wherein each set of data predicted coefficients for the respective classes is stored in a memory.

41. A hierarchical encoding method according to claim 40, wherein each set of data predicted coefficients for the respective classes is generated by using a method of least squares, and the predicted pixel generating step calculates a linear combination of a plurality of pixels of the second image signal and the set of predicted coefficients from the memory to generate the considered pixel.

42. A method for decoding a hierarchically encoded first image signal and producing a first image signal representing pixels and a first resolution, the first image signal being encoded by forming a second image signal and a differential signal from the first image signal, the second image signal having a second resolution lower than the first resolution, the differential signal being formed by generating a differential value between the pixel of the first image signal and predicted pixel of the first image signal produced from the second image signal, the method comprising the steps of:

receiving the second image signal and the differential signal;

producing predicted pixels of the first image signal from the second image signal; and

adding the predicted pixel of the second image and a corresponding differential value of the differential signal to producing the first image;

wherein the predicting step comprises the steps of;

forming predicted pixels for a pixel to be predicted of the second image signal from a plurality of pixels of the second image signal and each set of class predicted coefficients of respective classes from the memory, the plurality of pixels being disposed adjacent the pixel to be predicted of the second image signal;

determining a class from the predicted pixels and the pixel to be predicted as a class of a considered pixel of the first image signal;

generating a set of data predicted coefficients in response to the determined class, each set of data predicted coefficients for respective classes is formed by using a standard image signal having a resolution corresponding to the first resolution of the first image signal; and

generating a predicted pixel corresponding to the considered pixel from a plurality of pixels of the second image signal and the generated set of the data predicted coefficients.

43. A method according to claim 42, wherein each class predicted coefficients for the respective classes is generated by using a standard image signal having a resolution corresponding to the second resolution of the second image signal.

44. A method according to claim 43, wherein the predicted pixels forming step forms the predicted pixels for the pixel to be predicted of the second image signal by calculating a linear combination of the plurality of pixels of the second image signal and each set of class predicted coefficients of the respective classes from the memory;

and the class determining step determines a class corresponding to a set of class predicted coefficients to form the predicted pixel which is closest to the pixel to be predicted, in the formed predicted pixels, as a class of the considered pixel.

45. A method according to claim 42, wherein each set of data predicted coefficients for the respective classes is stored in a memory.

46. A method according to claim 45, wherein each set of data predicted coefficients for the respective classes is generated by using a method of least squares, and the predicted pixel generating step calculates a linear combination of a plurality of pixels of the second image signal and the set of predicted coefficients from the memory to generate the considered pixel.

47. A hierarchical encoding method for encoding a first image signal representing pixels and having a first resolution, the method comprising the steps of:

forming a second image signal from the first image signal, the second image signal having a second resolution lower than the first resolution;

producing predicted pixels of the first image signal from the second image signal;

generating a differential value between the pixel of the first image signal and the predicted pixel of the first image signal to a differential signal; and

transmitting the second image signal and the differential signal;

wherein the predicting step comprises the steps of;

forming predicted pixels for a pixel to be predicted of the second image signal from a plurality of pixels of the second image signal and each set of class predicted coefficients of respective classes from the memory, the plurality of pixels being disposed adjacent the pixel to be predicted of the second image signal;

determining a class from the predicted pixels and the pixel to be predicted as a class of a considered pixel of the first image signal; and

generating a predicted pixel corresponding to the considered pixel in response to the determined class, each predicted pixel for respective classes is formed by using a standard image signal having a resolution corresponding to the first resolution of the first image signal.

48. A hierarchical encoding method according to claim 47, wherein each class predicted coefficients for the respective classes is generated by using a standard image signal having a resolution corresponding to the second resolution of the second image signal.

49. A hierarchical encoding method according to claim 48, wherein the predicted pixels forming step forms the predicted pixels for the pixel to be predicted of the second image signal by calculating a linear combination of the plurality of pixels of the second image signal and each set of class predicted coefficients of the respective classes from the memory;

and the class determining step determines a class corresponding to a set of class predicted coefficients to form the predicted pixel which is closest to the pixel to be predicted, in the formed predicted pixels, as a class of the considered pixel.

50. A hierarchical encoding method according to claim 47, wherein each predicted pixel for the respective classes is stored in a memory.

51. A method for decoding a hierarchically encoded first image signal and producing a first image signal representing pixels and a first resolution, the first image signal being encoded by forming a second image signal and a differential signal from the first image signal, the second image signal having a second resolution lower than the first resolution, the differential signal being formed by generating a differential value between the pixel of the first image signal and predicted pixel of the first image signal produced from the second image signal, the method comprising the steps of:

receiving the second image signal and the differential signal;

producing predicted pixels of the first image signal from the second image signal; and

adding the predicted pixel of the second image signal and a corresponding differential value of the differential signal to producing the first image;

wherein the predicting step comprises the steps of;

forming predicted pixels for a pixel to be predicted of the second image signal from a plurality of pixels of the second image signal and each set of class predicted coefficients of respective classes from the memory, the plurality of pixels being disposed adjacent the pixel to be predicted of the second image signal;

determining a class from the predicted pixels and the pixel to be predicted as a class of a considered pixel of the first image signal; and

generating a predicted pixel corresponding to the considered pixel in response to the determined class, each predicted pixel for respective classes is formed by using a standard image signal having a resolution corresponding to the first resolution of the first image signal.

52. A method according to claim 51, wherein each class predicted coefficients for the respective classes is generated by using a standard image signal having a resolution corresponding to the second resolution of the second image signal.

53. A method according to claim 52, wherein the predicted pixels forming step forms the predicted pixels for the pixel to be predicted of the second image signal by calculating a linear combination of the plurality of pixels of the second image signal and each set of class predicted coefficients of the respective classes from the memory;

and the class determining step determines a class corresponding to a set of class predicted coefficients to form the predicted pixel which is closest to the pixel to be predicted, in the formed predicted pixels, as a class of the considered pixel.

54. A method according to claim 51, wherein each predicted pixel for the respective classes is stored in a memory.

55. A hierarchical encoding method for encoding a first image signal representing pixels and having a first resolution, the method comprising the steps of:

forming a second image signal from the first image signal, the second image signal having a second resolution lower than the first resolution;

producing predicted pixels of the first image signal from the second image signal;

generating a differential value between the pixel of the first image signal and the predicted pixel of the first image signal to a differential signal; and

transmitting the second image signal and the differential signal;

wherein the predicting step comprises the steps of;

forming predicted pixels for a pixel to be predicted of the second image signal from a plurality of pixels of the second image signal and each set of class predicted coefficients of respective classes from the memory, the plurality of pixels being disposed adjacent the pixel to be predicted of the second image signal;

determining a class from the predicted pixels and the pixel to be predicted as a class of a considered pixel of the first image signal;

generating a predicted data corresponding to the considered pixel in response to the determined class, each predicted data for respective classes is formed by using a standard image signal having a resolution corresponding to the first resolution of the first image signal; and

producing a predicted image signal corresponding to the first image signal from the predicted data.

56. A hierarchical encoding method according to claim 55, wherein each class predicted coefficients for the respective classes is generated by using a standard image signal having a resolution corresponding to the second resolution of the second image signal.

57. A hierarchical encoding method according to claim 56, wherein the predicted pixels forming step forms the predicted pixels for the pixel to be predicted of the second image signal by calculating a linear combination of the plurality of pixels of the second image signal and each set of class predicted coefficients of the respective classes from the memory;

and the class determining step determines a class corresponding to a set of class predicted coefficients to form the predicted pixel which is closest to the pixel to be predicted, in the formed predicted pixels, as a class of the considered pixel.

58. A hierarchical encoding method according to claim 57, wherein each predicted data for the respective classes is stored in a memory.

59. A method for decoding a hierarchically encoded first image signal and producing a first image signal representing pixels and a first resolution, the first image signal being encoded by forming a second image signal and a differential signal from the first image signal, the second image signal having a second resolution lower than the first resolution, the differential signal being formed by generating a differential value between the pixel of the first image signal and predicted pixel of the first image signal produced from the second image signal, the method comprising the steps of:

receiving the second image signal and the differential signal;

producing predicted pixels of the first image signal from the second image signal; and

adding the predicted pixel of the second image signal and a corresponding differential value of the differential signal to producing the first image; and

wherein the predicting step comprises the steps of;

forming predicted pixels for a pixel to be predicted of the second image signal from a plurality of pixels of the second image signal and each set of class predicted coefficients of respective classes from the memory, the plurality of pixels being disposed adjacent the pixel to be predicted of the second image signal;

determining a class from the predicted pixels and the pixel to be predicted as a class of a considered pixel of the first image signal;

generating a predicted data corresponding to the considered pixel in response to the determined class, each predicted data for respective classes is formed by using a standard image signal having a resolution corresponding to the first resolution of the first image signal; and

producing a predicted image signal corresponding to the first image signal from the predicted data.

60. A method according to claim 59, wherein each class predicted coefficients for the respective classes is generated by using a standard image signal having a resolution corresponding to the second resolution of the second image signal.

61. A method according to claim 60, wherein the predicted pixels forming step forms the predicted pixels for the pixel to be predicted of the second image signal by calculating a linear combination of the plurality of pixels of the second image signal and each set of class predicted coefficients of the respective classes from the memory;

and the class determining step determines a class corresponding to a set of class predicted coefficients to form the predicted pixel which is closest to the pixel to be predicted, in the formed predicted pixels, as a class of the considered pixel.

62. A method according to claim 59, wherein each predicted data for the respective classes is stored in a memory.