METHOD OF ENCODING DIGITAL SIGNALS (VARIANTS), METHOD OF DECODING DIGITAL SIGNALS (VARIANTS), DEVICE FOR IMPLEMENTATION THEREOF (VARIANTS), AND SYSTEM FOR IMAGE TRANSMISSION VIA LIMITED THROUGHPUT COMMUNICATION CHANNELS (VARIANTS)

• US 20100226569A1
• Filed: 09/28/2007
• Published: 09/09/2010
• Est. Priority Date: 09/28/2007
• Status: Active Grant

• Current Assignee: Sif Codec, LLC
• Original Assignee: Vsevolod Yurievich Mokrushin
• Inventors: Mokrushin, Vsevolod Yurievich
• Granted Patent: US 8,374,446 B2
• Time in Patent Office: Days
• US Class Current: 382/166
• CPC Class Codes: H04N 19/117 Filters, e.g. for pre-processing or post-processing sub-band filter banks H04N19/635
• CPC Class Codes: H04N 19/12 Selection from among a plurality of transforms or standards, e.g. selection between discrete cosine transform [DCT] and sub-band transform or selection between H.263 and H.264
• CPC Class Codes: H04N 19/33 in the spatial domain
• CPC Class Codes: H04N 19/46 Embedding additional information in the video signal during the compression process H04N19/517, H04N19/68, H04N19/70 take precedence
• CPC Class Codes: H04N 19/59 involving spatial sub-sampling or interpolation, e.g. alteration of picture size or resolution
• CPC Class Codes: H04N 19/635 characterised by filter definition or implementation details
• CPC Class Codes: H04N 19/80 Details of filtering operations specially adapted for video compression, e.g. for pixel interpolation H04N19/635, H04N19/86 take precedence

Abstract

The invention relates to systems for a transmission of images via channels of communication with a limited capacity by means of application of compression of the images. The technical result consists in an increase of the compression degree upon encoding, and it allows to fulfill the transmission of such encoded images via the channel of communication with the limited capacity, therewith a high degree of the compression is provided without of increase of computational power of encoding device and without of distortions upon decoding. The result is obtained by the usage of more effective method of interpolation of restored subsamples, in this method there is used an adaptive and applicative set of samples, which restore a quantized signal, and this set of the samples allows to improve an accuracy of the interpolation with a number of the subsamples, which is necessary for the right interpolation that is simultaneously decreased one.

Claims

1. Method of encoding digital signals, where, by means of computing facilities, as well as adaptive and/or linear overlapping low-pass filters, a hierarchical pyramid is formed based on encoded data, signal differences between this pyramid'"'"'s levels are calculated to be used in the course of preliminary classification of the aforesaid hierarchical pyramid, which is carried out successively, from the level of a section with the lowest spatial resolution to the level of a section with the highest spatial resolution, and the aforesaid hierarchical pyramid is finally classified with a predetermined set of resampling patterns, whereupon each section of the aforesaid hierarchical pyramid selected at the stage of preliminary classification is converted into an array of subsamples, which is achieved by means of adaptive and/or linear filters set in the resampling pattern put in line with the aforesaid section at the final classification, and the obtained array is converted into a one-dimensional data array, with information on location of obtained subsamples and resampling patterns used for obtaining thereof saved, whereupon the aforesaid array is encoded through lossy compression, and information on location of subsamples is encoded through lossless compression.

2. Method according to claim 1, characterized in that the calculated signal difference between the levels of the aforesaid hierarchical pyramid is used to change the compression threshold at different sections of encoded data.

3. Method according to claim 1, characterized in that, if no pattern can be selected from the predetermined set of patterns for a section of the aforesaid hierarchical pyramid chosen in the course of preliminary classification, samples contained in such a section are directly added to the array of subsamples to be created, whereas information is saved stating that none of the patterns was used on the section in question.

4. Method according to claim 1, characterized in that information on location of obtained subsamples and the type of resampling patterns used for obtaining thereof, which is saved in the course of encoding, is presented in the form of a hierarchical tree structure.

5. Method according to claim 1, characterized in that signal differences between the levels of the aforesaid hierarchical pyramid are calculated by means of adaptive and/or linear overlapping interpolation filters.

6. Method according to claim 1, characterized in that the process of compression of the one-dimensional array of subsamples is optimized by means of statistical data collected at the stages of preliminary and final classification.

7. Method according to claim 1, characterized in that preliminary and final classifications are optimized, with a criterion of minimization of the common distortion level used for a predetermined data level.

8. Method according to claim 1, characterized in that the process of converting an array of subsamples into a one-dimensional array is carried out as an adaptive process, its parameters changing depending on statistics of signals being processed.

9. Method according to claim 1, characterized in that a hierarchical pyramid is formed based on encoded data with at least three decimation levels.

10. Method according to claim 1, characterized in that in the course of preliminary classification of sections of the aforesaid hierarchical pyramid, sections of different fixed size are used, whereas in the course of final classification by means of a predetermined set of resampling patterns, several resampling patterns are used, each of them intended for its own size of section being processed.

11. Method according to claim 1, characterized in that contents, or a part of contents, of the used set of resampling patterns is encoded through the lossless method and saved together with encoded data.

12. Method according to claim 1, characterized in that data being encoded constitute full-color image presented as a number of color planes, each of which is encoded separately.

13. Method according to claim 1, characterized in that the process of encoding a one-dimensional data array is carried out through discrete cosine transform and linear and/or non-linear adaptive quantization, with subsequent encoding of data obtained after quantization, together with quantization parameters, through a method of encoding with variable-length codes.

14. Method according to claim 1, characterized in that the process of encoding a one-dimensional data array is carried out through stationary and/or adaptive wavelet transform, as well as linear and/or non-linear adaptive quantization, with subsequent encoding of data obtained after quantization, together with quantization parameters, through a method of encoding with variable-length codes.

15. Method according to claim 1, characterized in that the process of encoding a one-dimensional data array is carried out through stationary and/or adaptive sub-band transform, as well as linear and/or non-linear adaptive quantization, with subsequent encoding of data obtained after quantization, together with quantization parameters, through a method of encoding with variable-length codes.

16. Method according to claim 13, characterized in that, instead of the method of encoding with variable-length codes, the arithmetic encoding method is used, jointly with a predetermined and/or adaptive statistical model.

17. Method according to claim 1, characterized in that, before being encoded, the one-dimensional data array undergoes preliminary segmentation, whereas encoding is carried out through several various lossy compression methods to be selected individually for each obtained segment.

18. Method according to claim 4, characterized in that the method of encoding with variable-length codes is used for encoding information on location of subsamples and the type of resampling patterns applied.

19. Method according to claim 4, characterized in that the method of arithmetic encoding jointly with a predetermined and/or adaptive statistical model is used for encoding information on location of subsamples and the type of resampling patterns applied.

20. Method according to claim 1, characterized in that its operations are performed for an entire array of data being encoded.

21. Method according to claim 1, characterized in that the array of data being encoded is preliminarily divided into overlapping blocks, with operations performed for each block independently from the others.

63. Device for implementation of the method of encoding digital signals according to claim 1, which comprises a unit of forming and storing hierarchical pyramid from input data, a unit of calculating interlevel difference, a unit of preliminary classification of data sections, a unit of final classification, filtration and resampling, a unit of converting filtered subsamples into a one-dimensional array, a unit of encoding location of subsamples and lossless compression, a unit of lossy compression of one-dimensional array, an output driver unit, and a storage of a set of predetermined resampling patterns, where the unit of forming and storing hierarchical pyramid from input data is connected to the unit of calculating interlevel difference and the unit of final classification, filtration and resampling, whereas the unit of calculating interlevel difference is connected to the unit of preliminary classification of data sections, where the latter one is connected to the unit of final classification, filtration and resampling, which, in its turn, is connected to the storage of a set of predetermined resampling patterns and to the unit of converting filtered subsamples into a one-dimensional array, which is connected to the unit of encoding location of subsamples and lossless compression, and to the unit of lossy compression of one-dimensional array, their outputs connected to the output driver unit.

64. Device according to claim 63, characterized in that it comprises a unit of calculating local compression thresholds connected to the unit of calculating interlevel difference and to the unit of preliminary classification of data sections.

65. Device according to claim 63, characterized in that the unit of preliminary classification of data section classifies the data into sections of different fixed size, while the unit of final classification, filtration and resampling processes sections of different fixed size

66. Device according to claim 63, characterized in that it comprises a unit of lossless encoding of resampling patterns, which is connected to the storage of a set of predetermined resampling patterns, and to the output driver unit.

67. Device according to claim 63, characterized in that the process of encoding a full-color image comprising a number of color planes is performed independently for each of the aforesaid planes.

68. Device according to claim 63, characterized in that the output driver unit adds more information to encoded data.

22. Method of decoding digital signals presented as a one-dimensional array of data obtained from subsamples taken from sections of a hierarchical pyramid formed from digital signals being encoded (the aforesaid array compressed through the lossy compression method) and information on location of these subsamples in the aforesaid pyramid encoded through the lossless compression method, where there is performed restoration of the array of subsamples through the lossy decoding method, decoding information on location of transmitted subsamples through the lossless method, restoration of transmitted sections of the hierarchical pyramid with the restored array of subsamples and decoded information on their location, restoration of missed sections of the hierarchical pyramid, for which there are restored similarly located sections on levels with a higher spatial resolution, performed by means of non-overlapping filters with subsequent decimation, whereupon the process of final interpolation of transmitted data is performed with adaptive and/or linear composite cascade overlapping interpolation filters, which is carried out successively, from the level of a section with the lowest spatial resolution to the level of a section of the aforesaid pyramid with the highest spatial resolution, whereupon restored digital signals are read from the hierarchical pyramid level with the highest spatial resolution.

23. Method according to claim 22, characterized in that the encoded one-dimensional data array is formed of subsamples obtained from sections of the aforesaid hierarchical pyramid by means of adaptive and/or linear filters set in the resampling pattern used in encoding, the above pattern selected out of a predetermined set of resampling patterns, where information on the type of resampling patterns applied is encoded together with information on location of transmitted subsamples;

here, in the course of restoration of transmitted sections of the hierarchical pyramid, non-overlapping parts of adaptive and/or linear composite cascade interpolation filters are used, which are determined for each sample belonging to a section being restored of the hierarchical pyramid in a predetermined set of resampling patterns, whereas, concurrently with the process of final interpolation, there is performed final filtration of transmitted data sections with overlapping compensating parts of adaptive and/or linear composite cascade interpolation filters determined for each sample belonging to a section being restored of the hierarchical pyramid in a predetermined set of resampling patterns.

24. Method according to claim 22, characterized in that the process of restoration of missed sections of the hierarchical pyramid, for which there are restored similarly located sections on levels with a higher spatial resolution, is performed without using non-overlapping low-pass filters;

instead, direct decimation is performed for sections with a higher spatial resolution.

25. Method according to claim 22, characterized in that, the transmitted data sections having been restored, the process of restoration is performed for the hierarchical pyramid sections missed in the course of restoration, for which there are restored similarly located sections on the pyramid levels with a lower spatial resolution, which is performed by means of non-overlapping interpolation filters, whereas the subsequent restoration of the missed sections of the hierarchical pyramid, for which there are restored similarly located sections on the levels with a higher spatial resolution, is performed by means of overlapping adaptive and/or linear filters with subsequent decimation.

26. Method according to claim 25, characterized in that an additional stage of restoration of missed sections and the corresponding stage of restoration by means of overlapping adaptive and/or linear filters with subsequent decimation are only performed for parts of the levels of the hierarchical pyramid being restored, which have low spatial resolution.

27. Method according to claim 23, characterized in that the aforesaid subsamples are obtained from sections of the above hierarchical pyramid, which have different fixed sizes, whereas the processes of restoration of transmitted data sections and final filtration are performed by means of several sets of resampling patterns, each of them intended for its own fixed section size.

28. Method according to claim 22, characterized in that encoded data contain information encoded through the lossless method, which comprises resampling patterns used in the course of encoding, or just parts of these patterns;

here, restoration of encoded data is preceded by decoding of the aforesaid information, whereupon decoded patterns, or parts thereof, are added to the set of resampling patterns used in the process of restoration of encoded data.

29. Method according to claim 22, characterized in that encoded data constitute a number of separately encoded color planes belonging to a full-color image, where the process of decoding is performed independently for each of the encoded color planes.

30. Method according to claim 22, characterized in that its operations are performed for an entire array of data being restored.

31. Method according to claim 22, characterized in that the array of data being restored is preliminarily divided into overlapping blocks, with operations performed for each block independently from the others.

69. Device for implementation of the method for decoding digital signals according to claim 23, which contains a unit of receiving input encoded data and demultiplexing, a unit of lossless decompression and decoding of information on location of subsamples, a unit of decoding a one-dimensional array of subsamples encoded by means of a lossy compression algorithm, a unit of restoration of transmitted data sections, a unit of restoration of missed sections, for which there are similar restored sections on the pyramid levels with a higher spatial resolution, a unit of final filtration and interpolation, a unit of storing and output of hierarchical pyramid, a storage of a set of predetermined resampling patterns;

where the unit of receiving input encoded data and demultiplexing is connected to the unit of lossless decompression and decoding of information on location of subsamples, and to the unit of restoration of a one-dimensional array of subsamples encoded by means of a lossy compression, their outputs connected to the unit of restoration of transmitted data sections, which is connected to the unit of storing and output of hierarchical pyramid, the storage of a set of predetermined resampling patterns and the unit of restoration of missed sections, for which there are similar restored sections on the pyramid levels with a higher spatial resolution, the latter being connected to the unit of storing and output of hierarchical pyramid and the unit of final filtration and interpolation, which is connected to the unit of storing and output of hierarchical pyramid, and the storage of a set of predetermined resampling patterns.

70. Device according to claim 69, characterized in that it comprises a unit of interpolation of missed sections, for which there are restored similarly located sections on the levels with a lower spatial resolution, which is connected to the opening of the link between the unit of restoration of transmitted data sections and the unit of restoration of missed sections, for which there are similar restored sections on the pyramid levels with a higher spatial resolution, and also connected to the unit of storing and output of hierarchical pyramid.

71. Device according to claim 69, characterized in that the unit of restoration of transmitted data sections and the unit of final filtration and interpolation work with data sections of different fixed sizes, while using several sets of resampling patterns stored in the storage of a set of resampling patterns, where each of the aforesaid sets is intended for its own fixed size of the section being processed.

72. Device according to claim 69, characterized in that it comprises a unit of lossless decoding of transmitted patterns, which is connected to the unit of receiving input encoded data and demultiplexing, and to the storage of a set of predetermined resampling patterns.

73. Device according to claim 69, characterized in that decoding of encoded full-color image comprising a number of color planes is performed independently for each of the aforesaid planes.

74. Device according to claim 69, characterized in that the unit of receiving encoded data and demultiplexing singles out additional information from encoded data.

32. Method of encoding digital signals, where, by means of computing facilities, as well as adaptive and/or linear overlapping low-pass filters, a hierarchical pyramid is formed based on encoded data, as well as reference hierarchical pyramids of the same size based on reference data arrays, whereupon, based on the reference hierarchical pyramids and the original hierarchical pyramid, differential hierarchical pyramids are calculated, whereupon signal differences are calculated between the levels of the hierarchical pyramid created based on encoded data and the levels of the aforesaid differential pyramids, with the resulting interlevel difference to be used in the process of preliminary classification of the aforesaid hierarchical pyramid and obtained differential hierarchical pyramids, which is performed successively, from the level of a section with the lowest spatial resolution to the level of a section with the highest spatial resolution, where the aforesaid hierarchical pyramid and the obtained differential hierarchical pyramids are finally classified with a predetermined set of resampling patterns, whereupon each section of the aforesaid hierarchical pyramid and the section of the differential pyramids selected at the stage of preliminary classification is converted into an array of subsamples, which is achieved by means of adaptive and/or linear filters set in the resampling pattern put in line with the aforesaid section at the final classification, and the obtained subsamples are converted into at least two data arrays, for subsamples obtained from the aforesaid pyramid, and for subsamples obtained from differential pyramids, with information on location of obtained subsamples and resampling patterns used for obtaining thereof saved, whereupon the aforesaid arrays are encoded through lossy compression, and information on location of subsamples is encoded through lossless compression.

33. Method according to claim 32, characterized in that the calculated signal difference between the levels of the aforesaid hierarchical pyramid and the levels of the differential pyramids is used to change the compression threshold at different sections of encoded data.

34. Method according to claim 32, characterized in that, if no pattern can be selected from the predetermined set of patterns for a section of the aforesaid hierarchical pyramid and the differential pyramids chosen in the course of preliminary classification, samples contained in such a section are directly added to the array of subsamples to be created, whereas information is saved stating that none of the patterns was used on the section in question.

35. Method according to claim 32, characterized in that information on location of obtained subsamples and the type of resampling patterns used for obtaining thereof, which is saved in the course of encoding, is presented in the form of a hierarchical tree structure.

36. Method according to claim 32, characterized in that signal differences between the levels of the aforesaid hierarchical pyramid and between the levels of the differential pyramids are calculated by means of adaptive and/or linear overlapping interpolation filters.

37. Method according to claim 32, characterized in that the process of compression of one-dimensional arrays of subsamples is optimized by means of statistical data collected at the stages of preliminary and final classification.

38. Method according to claim 32, characterized in that preliminary and final classifications are optimized, with a criterion of minimization of the common distortion level used for a predetermined data level.

39. Method according to claim 32, characterized in that the process of converting an array of subsamples into one-dimensional arrays is carried out as an adaptive process, its parameters changing depending on statistics of signals being processed.

40. Method according to claim 32, characterized in that a hierarchical pyramid is formed based on encoded data, and reference pyramids are formed based on reference data, with at least three decimation levels.

41. Method according to claim 32, characterized in that, in the course of preliminary classification of sections of the aforesaid hierarchical pyramid and differential hierarchical pyramids, sections of different fixed size are used, whereas in the course of final classification by means of a predetermined set of resampling patterns, several resampling patterns are used, each of them intended for its own size of section being processed.

42. Method according to claim 32, characterized in that contents, or a part of contents, of the used set of resampling patterns is encoded through the lossless method and saved together with encoded data.

43. Method according to claim 32, characterized in that encoded data constitute a full-color image presented as a number of color planes, each of which is encoded separately.

44. Method according to claim 32, characterized in that the process of encoding one-dimensional data arrays is carried out through discrete cosine transform and linear and/or non-linear adaptive quantization, with subsequent encoding of data obtained after quantization, together with quantization parameters, through a method of encoding with variable-length codes.

45. Method according to claim 32, characterized in that the process of encoding one-dimensional data arrays is carried out through stationary and/or adaptive wavelet transform, as well as linear and/or non-linear adaptive quantization, with subsequent encoding of data obtained after quantization, together with quantization parameters, through a method of encoding with variable-length codes.

46. Method according to claim 32, characterized in that the process of encoding one-dimensional data arrays is carried out through stationary and/or adaptive sub-band transform, as well as linear and/or non-linear adaptive quantization, with subsequent encoding of data obtained after quantization, together with quantization parameters, through a method of encoding with variable-length codes.

47. Method according to claim 44, characterized in that, instead of the method of encoding with variable-length codes, the arithmetic encoding method is used, jointly with a predetermined and/or adaptive statistical model.

48. Method according to claim 32, characterized in that, before being encoded, one-dimensional data arrays undergo preliminary segmentation, whereas encoding is carried out through several various lossy compression methods to be selected individually for each obtained segment.

49. Method according to claim 35, characterized in that method of encoding with variable-length codes is used for encoding information on location of subsamples and the type of resampling patterns applied.

50. Method according to claim 35, characterized in that the method of arithmetic encoding, jointly with a predetermined and/or adaptive statistical model, is used for encoding information on location of subsamples and the type of resampling patterns applied.

51. Method according to claim 32, characterized in that its operations are performed for an entire array of data being encoded.

52. Method according to claim 32, characterized in that the array of data being encoded is preliminarily divided into overlapping blocks, with operations performed for each block independently from the others.

75. Device for implementation of the method of encoding digital signals according to claim 32, which comprises a unit of forming and storing a hierarchical pyramid from input data, a unit of forming reference hierarchical pyramids from reference data, a unit of calculating and storing differential hierarchical pyramids, a unit of calculating interlevel difference, a unit of preliminary classification of data sections, as well as a unit of final classification, filtration and resampling, a unit of converting filtered subsamples into one-dimensional arrays, a unit of encoding location of subsamples and lossless compression, a unit of lossy compression of one-dimensional data arrays, a unit of output driver and a storage of a set of predetermined resampling patterns, where the unit of forming and storing hierarchical pyramid from input data is connected to the unit of calculating and storing differential hierarchical pyramids, as well as to the unit of calculating interlevel difference and the unit of final classification, filtration and resampling, while the unit of forming reference hierarchical pyramids from reference data is connected to the unit of calculating and storing differential hierarchical pyramids, whereas the latter one is connected to the unit of final classification, filtration and resampling, and to the unit of calculating interlevel difference, where the latter one is connected to the unit of preliminary classification of data sections, which is connected to the unit of final classification, filtration and resampling, which, in its turn, is connected to the storage of a set of predetermined resampling patterns, and to the unit of converting filtered subsamples into one-dimensional arrays, which is connected to the unit of encoding location of subsamples and lossless compression, and to the unit of lossy compression of data arrays, their outputs connected to the output driver unit.

76. Device according to claim 75, characterized in that it comprises a unit of calculating local compression thresholds, which is connected to the unit of calculating interlevel difference, and to the unit of preliminary classification of data sections.

77. Device according to claim 75, characterized in that the unit of preliminary classification of data sections classifies the data into sections of different fixed size, while the unit of final classification, filtration and resampling processes sections of different fixed size.

78. Device according to claim 75, characterized in that it comprises a unit of lossless encoding of resampling patterns, which is connected to the storage of a set of predetermined resampling patterns, and to the output driver unit.

79. Device according to claim 75, characterized in that the process of encoding a full-color image comprising a number of color planes is performed independently for each of the aforesaid planes.

80. Device according to claim 75, characterized in that the output driver unit adds more information to encoded data.

53. Method of decoding digital signals presented as one-dimensional arrays of data obtained from subsamples taken from sections of a hierarchical pyramid formed from digital signals being encoded and sections of the differential pyramids obtained from the aforesaid pyramid and from reference hierarchical pyramids of the same size obtained-from reference data arrays (the aforesaid one-dimensional arrays compressed through the lossy compression method) and information on location of these subsamples in the aforesaid pyramid and in the differential pyramids encoded through the lossless compression method, lies in the fact that there is performed formation of reference hierarchical pyramids based on reference data arrays, decoding of information describing location of transmitted subsamples in the hierarchical pyramid based on encoded data, as well in differential pyramids, which is performed through a lossless decompression method, restoration of the arrays of subsamples through the lossy decoding method, whereupon there is performed restoration of transmitted sections of the aforesaid hierarchical pyramid with the restored arrays of subsamples and decoded information on their location, as well as with reference hierarchical pyramids for sections being restored from subsamples obtained from differential pyramids, whereupon there is performed restoration of missed sections of the hierarchical pyramid, for which there are restored similarly located sections on levels with a higher spatial resolution, performed by means of non-overlapping filters with subsequent decimation, whereupon the process of final interpolation of transmitted data is performed with adaptive and/or linear composite cascade overlapping interpolation filters, which is carried out successively, from the level of a section with the lowest spatial resolution to the level of a section with the highest spatial resolution, where for sections of the aforesaid pyramid restored from subsamples obtained from differential pyramids, and for sections obtained from those sections by means of interpolation, the processes of final filtration and interpolation are performed with using the relevant sections of reference hierarchical pyramids, whereupon restored digital signals are read from the hierarchical pyramid level with the highest spatial resolution.

54. Method according to claim 53, characterized in that encoded one-dimensional data arrays are formed of subsamples obtained from sections of the aforesaid hierarchical pyramid and sections of the differential pyramids by means of adaptive and/or linear filters set in the resampling pattern used in encoding, the above pattern selected out of a predetermined set of resampling patterns, where information on the type of resampling patterns applied is encoded together with information on location of transmitted subsamples;

here, in the course of restoration of transmitted sections of the hierarchical pyramid, non-overlapping parts of adaptive and/or linear composite cascade interpolation filters are used, which are determined for each sample belonging to a section being restored of the hierarchical pyramid in a predetermined set of resampling patterns, whereas, concurrently with the process of final interpolation, there is performed final filtration of transmitted data sections with overlapping compensating parts of adaptive and/or linear composite cascade interpolation filters determined for each sample belonging to a section being restored of the hierarchical pyramid in a predetermined set of resampling patterns, where, for sections of the aforesaid pyramid restored from subsamples obtained from differential pyramids, all the above-mentioned processes of filtration and interpolation are performed with using the relevant sections of the reference hierarchical pyramids.

55. Method according to claim 53, characterized in that the process of restoration of missed sections of the hierarchical pyramid, for which there are restored similarly located sections on levels with a higher spatial resolution, is performed without using non-overlapping low-pass filters;

instead, direct decimation is performed for sections with a higher spatial resolution.

56. Method according to claim 53, characterized in that, the transmitted data sections having been restored, the process of restoration is performed for the hierarchical pyramid sections missed in the course of restoration, provided that there are restored similarly located sections on the pyramid levels with a lower spatial resolution, which is performed by means of non-overlapping interpolation filters, where, for sections of the aforesaid pyramid restored from subsamples obtained from differential pyramids, the aforesaid interpolation is performed with using the relevant sections of the reference hierarchical pyramids;

besides, subsequent restoration of the missed sections of the hierarchical pyramid, for which there are restored similarly located sections on the levels with a higher spatial resolution, is performed by means of overlapping adaptive and/or linear filters with subsequent decimation.

57. Method according to claim 56, characterized in that an additional stage of restoration of missed sections and the corresponding stage of restoration by means of overlapping adaptive and/or linear filters with subsequent decimation are only performed for the part of the levels of the hierarchical pyramid being restored, which have low spatial resolution.

58. Method according to claim 54, characterized in that the aforesaid subsamples are obtained from sections of the above hierarchical pyramid, which have different fixed sizes, whereas the processes of restoration of transmitted data sections and final filtration are performed by means of several sets of resampling patterns, each of them intended for its own fixed section size.

59. Method according to claim 53, characterized in that encoded data contain information encoded through the lossless method, which comprises resampling patterns used in the course of encoding, or just parts of these patterns;

here, restoration of encoded data is preceded by decoding of the aforesaid information, whereupon decoded patterns, or parts thereof, are added to the set of resampling patterns used in the process of restoration of encoded data.

60. Method according to claim 53, characterized in that encoded data constitute a number of separately encoded color planes belonging to a full-color image, where the process of decoding is performed independently for each of the encoded color planes.

61. Method according to claim 53, characterized in that its operations are performed for an entire array of data being restored.

62. Method according to claim 53, characterized in that the array of data being restored is preliminarily divided into overlapping blocks, with operations performed for each block independently from the others.

81. Device for implementation of the method for decoding digital signals according to claim 54, which comprises a unit of receiving input encoded data and demultiplexing, a unit of lossless decompression and decoding of information on location of subsamples, a unit of decoding one-dimensional arrays of subsamples encoded by means of a lossy compression algorithm, a unit of restoration of transmitted data sections, a unit of restoration of missed sections, for which there are similar restored sections on the pyramid levels with a higher spatial resolution, a unit of final filtration and interpolation, a unit of storing and output of hierarchical pyramid, a unit of forming and storing reference hierarchical pyramids from reference data, a storage of a set of predetermined resampling patterns, where the unit of receiving input encoded data and demultiplexing is connected to the unit of lossless decompression and decoding of information on location of subsamples, and to the unit of restoration of one-dimensional arrays of subsamples encoded by means of a lossy compression, their outputs connected to the unit of restoration of transmitted data sections, which is connected to the unit of storing and output of hierarchical pyramid, to the unit of forming and storing reference hierarchical pyramids from reference data, to the storage of a set of predetermined resampling patterns and the unit of restoration of missed sections, for which there are similar restored sections on the pyramid levels with a higher spatial resolution, the latter being connected to the unit of storing and output of hierarchical pyramid and to the unit of final filtration and interpolation, which is connected to the unit of storing and output of hierarchical pyramid, to the unit of forming and storing reference hierarchical pyramids from reference data, and to the storage of a set of predetermined resampling patterns.

82. Device according to claim 81, characterized in that it comprises a unit of interpolation of missed sections, for which there are restored similarly located sections on the levels with a lower spatial resolution, which is connected to the opening of the link between the unit of restoration of transmitted data sections and the unit of restoration of missed sections, for which there are similar restored sections on the pyramid levels with a higher spatial resolution, and also connected to the unit of storing and output of hierarchical pyramid, and to the unit of forming and storing reference hierarchical pyramids from reference data.

83. Device according to claim 81, characterized in that the unit of restoration of transmitted data sections and the unit of final filtration and interpolation work with data sections of different fixed sizes, while using several sets of resampling patterns stored in the storage of a set of resampling patterns, where each of the aforesaid sets is intended for its own fixed size of the section being processed.

84. Device according to claim 81, characterized in that it comprises a unit of lossless decoding of transmitted patterns, which is connected to the unit of receiving input encoded data and demultiplexing, and to the storage of a set of predetermined resampling patterns.

85. Device according to claim 81, characterized in that decoding of an encoded full-color image comprising a number of color planes is performed independently for each of the aforesaid planes.

86. Device according to claim 81, characterized in that the unit of receiving encoded data and demultiplexing singles out additional information from encoded data.

87. (canceled)

88. (canceled)