WO2004023820A1

WO2004023820A1 - A device and a system for providing a change of a video signal

Info

Publication number: WO2004023820A1
Application number: PCT/SE2003/001378
Authority: WO
Inventors: Harald Brusewitz
Original assignee: Iqc Transcoding Ab
Priority date: 2002-09-04
Filing date: 2003-09-04
Publication date: 2004-03-18
Also published as: SE525615C2; SE0202614D0; SE0202614L; AU2003256213A1

Abstract

A device and a system for converting or reducing the bit rate in a digital video signal, or for compressing a digital video signal. The video signal is subjected to a transformation utilising quantised transform components, and supporting information. The device as a separate one or included in the system may include a decoder, which receives the video signal to be converted, and a coder, which provides the digital signal in converted or reduced form. Alternatively, the device forms a coder. The respective device is arranged to modify solely the transform components, which means that the supporting information is at least substantially not influenced. The device enables a decrease of the bit rate in a coded video or compression of the video signal without the occurrence of any substantial deterioration of the image or signal quality.

Description

A device and a system for providing a change of a video signal

The present invention refers to a device for providing a video signal, which is transformed by quantised transform com- ponents, e.g. lev and/or run, and contains or can be supplied with supporting information, e.g. image type, macro- block, movement vectors, etc.

The invention also refers to a system constructed by or including one or several such devices.

Video signals may be changed or influenced, e.g. digitally, by various methods for conversion, reduction and compression, and the present invention is specifically directed to the case where transform technique is included. A common example of transform that hereby may be used is the Discrete Cosines Transform (DCT). The invention is however applicable on other types of transforms. Common for actual compression methods is that the transform components are quantised and that a variable length code (VLC) is used for creating code words representing the quantised value (frequently called "lev") of the transform component. In most compression algorithms also so called run- length coding is used, which means that a particular code word is used for representing the number of transform components (frequently called "run") that is quantised to the value zero (0) prior to a transform component having a value differing from zero. Several compression methods for video signals use a VLC where each code word represents a combination of the value of "run" and "lev". Examples of such compressions methods where this is valid are the standards (1 -8), inter alia JPEG and MPEG.

A straight and well known method of changing the bit rate in a coded bit stream is firstly to decode the bit stream and then to code the resulting video signal by a new bit rate, compare the following Fig 1 . In a more efficient technique, the information in the bit stream, which becomes known during the decoding procedure, is used. This information is less complex and may be transmitted via a side channel (compare the following Fig 2) to the coder, and may thus be used for simplifying the coder and for improving the performance of the rate conversion. Said side information may consist of image type, macro-block type, movement vectors etc. (Compare the standards mentioned above for image compression for definition of these items). An important property in this method is to save bits there through. This re-quantisation involves a deterioration of the image quality when the resulting bit stream later on is decoded. In a method equivalent with regard to the performance, the decoder and the following coder have been integrated in one loop. This method is described in scientific literature.

It is also generally referred to:

1 . ISO/IEC (JPEG)

2. ITU-T H.261

3. ITU-T H.262 4. ITU-T H.263

5. ITU-T-H.264

6. ISO/IEC 1 1 172-2 (MPPEG-1 )

7. ISO/IEC 13818-2 (MPPEG-2)

8. ISO/IEC 14496-2 (MPPEG-4) 9. "Transcoding of MEG bitstreams" Signal Processing:

Image Communication 8 (1996) pp 481 -500 G. Keesman et al.

10. "Real-time Transcoding of MPEG-2 Video Bit Streams"

International Broadcasting (IBC 97) PN Rudor and OH Werner (BBC).

In connection to devices and systems of relevant types express desiderata are inter alia present saying that the bit rate could be decreased substantially in a coded video, so-called rate conversion, without deteriorating substantially the content in an actual video signal, i. e. without the image perception is experienced as deteriorated or substantially better than when using known technique. There is also a desideratum of reducing substantially the number of bits or the bandwidth from an already coded or compressed still image or video signal without said deterioration occurring. The object of the invention is to solve inter alia the whole or parts of these problems.

There is also a need of generally improving the performance of coders of the relevant types. There is also a desideratum of easily being able to vary the parameters used in the device and the system. Thus the value of "lev" in certain selected DCT- components should, for instance, be possible to change without the need of re-quantisation. In an embodiment a feedback loop should be possible to use for avoiding error propagation from image to image. Such a loop should be possible to avoid if intracoding is used in the incoming (and thus the outgoing) bit stream. There is also a desideratum of being able to combine the features characteristic for the invention with re-quantisation. Instead of transmitting the same value of "quant" for each macro-block, the values "lev" of the transform components should alternatively be possible to change individually according to a particular criterion. This criteria should possibly be based on the fact that a large bit saving can be made at the same time as a resulting loss in the image quality is limited. There is also a desideratum of being able to reuse a part of the information in the incoming bit stream in the outgoing bit stream, and of being able to operate the device or the system with or without loop in the video case. The invention intends to solve also the whole or parts of this complex of problems.

In certain critical cases there may be a need of modifying other information in the bit stream than the transform components. Such a case may arise when the incoming bit stream contains errors, for instance due to poor receiving conditions. An erroneous bit stream may contain forbidden values of the side information, and it may then be suitable, for instance, to assign in a manner known per se allowable values to the erroneous parameters. The invention also solves this problem.

There is also a desideratum of being able to change the value or values of quantised DCT-components in a so-called rate converter or in a coder depending on the use of the device or the system. The invention also solves this problem.

The feature which may be regarded as essentially characterising a device according to the invention is that it is arranged to modify, during the change of the video signal, substantially solely the transform components in one of the following two cases, namely where the device forms a unit converting or reducing the video signal includes a decoder, preferably a VLD, which receives the video signal, and a coder, preferably a VLC, which effectuates a change in the front of a conversion or reduction of the video signal received, wherein said modification results in at least substantially no influence on the supporting information, and where the device forms a coder receiving the video signal arranged to provide by itself a change in the form of a compression of the video signal received and to generate the supporting information substantially independently of the compression.

In a further development of the inventive thought also the video buffer verification (vbv)-delay in the PIC-header is subjected to a change function. Also macro-block type and movement vectors may be subjected to change functions, which is valid in the case that the incoming bit stream is erroneous. The device may operate in a circuit arrangement with or without feedback circuit. When using a feedback circuit, the latter is provided to avoid error propagation from image to image. When modifying the transform component or transform components, the device may operate with or without re-quantisation. The device may also be provided to permit change of the value of one or several transform components, e. g. lev, according to a selected criteria, e. g. a criteria based on a large bit saving with simultaneous small reduction or substantially maintained image quality.

According to further embodiments, the device is arranged not to modify any transform component in an individual DCT-block, to modify an individual transform component in an individual DCT- block, to modify several transform components, but not all, in an individual DCT-block, to modify all transform components in an individual DCT-block, and/or to modify all transform components in an individual macro-block. Furthermore, an individual transform component, e.g. lev, may be changeable by one unit, e.g. from 1 (one) to 0 (zero) or e.g. from -1 (minus one) to 0 (zero) or e.g. from 4 (four) to 3 (three), in the case this involves a bit saving by a predetermined minimum number of bits. Alternatively, an individual transform component, e.g. lev, may be changeable by more than one unit, e.g. from 2 (two) to 0 (zero) or e.g. from -3 (minus three) to 0 (zero) or e.g. 4 (four) to 2 (two), if this involves a bit saving with a predetermined minimum number of bits. According to a further alternative embodiment, several transform components, e. g. lev, may together be changeable by one or several units if this involves a total bit-saving by a predetermined minimum number of bits.

A system according to the invention may essentially be characterised in that the respective device for providing an influence or change of the video signal either includes a decoder or a coder or itself forms a coder which compresses the received video signal, and that one or several of said devices are provided to modify substantially solely the transform components and that the supporting information thus is substantially not influenced. The respective device concerned may be included in or connectable to any of the following function alternative, recording on digital storing media for providing a memory saving and/or a modified image quality, relay transmission in or to a transmission network with smaller bandwidth, multiplexing of two or several video channels, transfer from variable to constant bit rate, transfer from constant to variable bit rate, transfer from constant to constant lower bit rate, transfer from variable to variable lower bit rate, or compression of the video signal between transmission and receiving equipment.

By the proposals identified above, an advantageous solution to the problems initially defined is obtained. The bit rate may, with a substantially maintained image quality, be decreased in an advantageous manner and the construction may to a large extent be based on known and technically simple means. Functions and parameters may be optimised and varied by means of program and computer controls, for instance via tele and/or data communication connections, also Internet. The invention may if so desired be based on standards and known techniques. The selection of criteria or parameters for determining or adjusting values of the transform components and thus the bit-saving and image quality may be performed by means of buffer control, for instance provided by a module. The invention also enables device and system constructions with less complexity than previously. Certain bits in the incoming bit stream may be used in the outgoing bit stream. No loop is required in the video case and the value of "lev" in certain selected DCT-components may be changed without the need of re-quantisation.

An embodiment, presently proposed, of a device and a system according to the invention is now to be described in the following with simultaneous reference to the drawings attached.

Fig 1 discloses in a block diagram and principally conversion of the bit rate by a decoder and coder according to known technique.

Fig 2 discloses in a block diagram and principally conversion of the bit rate by a decoder, a coder and supplementary information according to known technique. discloses in a block diagram and principally a decoder followed by a coder for still image according to known technique.

discloses in a block diagram and principally a decoder followed by a coder for video integrated in a simplified loop according to known technique.

discloses in a block diagram and principally the bit rate conversion according to the present invention without loop and without re-quantisation.

discloses in a block diagram and principally the bit rate conversion according to the present invention without loop and with re-quantisation.

discloses in a block diagram and principally the bit rate conversion according to the invention with loop and with re-quantisation.

discloses in a block diagram and principally a video coder according to known technique.

discloses in a block diagram and principally a video coder with modified DCT-coefficients according to the invention.

discloses in a block diagram and principally examples of recording on digital storing media with the use of the present invention.

discloses in a block diagram and principally examples of relay transmission to transmission networks with smaller bandwidth and with the use of the present invention.

discloses in a block diagram and principally an example of multiplexing of several video channels with the use of the present invention. Fig 1 defines symbolically a decoder 1 , a coder 2 and an incoming coded video signal 3 with a certain bit rate. A decoded video signal 4 is delivered from the decoder to the coder 2. After processing in the coder 2 a coded video signal 5 with a lower bit rate than the first mentioned bit rate is obtained at the output of the coder.

Fig 2 corresponds to the case according to Fig 1 with the difference that supplementary information, in addition to the uncoded information, is transferred between the decoder and the coder. The decoder is in this case defined by 6 and the coder by 7. A coded video signal has the reference sign 8 and a decoded, video signal has been defined by 9. By 10 is referred to a coded video signal with a lower bit rate than the one related to in Fig 1 . 1 1 refers to side information or a side information signal.

Fig 3 illustrates more in detail a case corresponding to Fig 1 for a still image. A variable length decoder is disclosed by 12 and an inverse quantiser 1 by 13. An Inverse Discrete Cosinus Transform (IDCT) 14 is connected to the latter component. 15 refers to a Discrete Cosinus Transform and 16 to a quantiser 2. A variable length coder is defined by 17. In the figure, the components are also defined by their respective abbreviations. To the component 12 a coded still image 18 is supplied and the component is outputting quantised DCT-components 19. At the output of the component, inverse quantised DCT-components 20 are obtained. 21 intends to symbolise a decoded still image supplied to the component 15. This delivers re-quantised DCT- components 22 to the component 16 which in its turn provides re-quantised DCT-components 23. 24 intends to illustrate a recorded still image with fewer bits than the coded still image in point 18.

In Fig 4, the embodiment according to Fig 2 has been replaced by a simplified loop. An inverse quantiser 28 has been connected to the output of the quantiser 27 and connected to a subtracter which is also connected to the incoming signal to the quantiser 27. The output of the subtracter is connected to an inverse transformer (IDCT = Inverse Discrete Cosine Transform). In this case a variable length decoder VLD is defined by 25, and an inverse quantiser 1 by IQι which has the reference sign 26. A quantiser 2 (Q₂) is defined by 27 and an inverse quantiser IQ₂ by 28. 29 refers to an Inverse Discrete Cosinus Transform (IDCT) 29. 30 is a movement compensator using a part of the side information, which movement compensator has the reference sign 30. To the movement compensator, a Discrete Cosinus Transform (DCT) is connected, which has the reference sign 31 . A variable length coder VLC is defined by 32. A coded video signal has the reference sign 33 and quantised DCT-components 34. Inverse quantised DCT- components have the reference sign 35 and DCT-components compensated for quantising errors in previous images, via a subtracter, have the reference sign 36. Re-quantised DCT- components have the reference sign 37 and inverse re- quantised DCT-components have been defined by 38. Quantising errors of DCT-components are defined by 39 and 40 intends to disclose a feedback inverse transformed quantising error. 41 is a movement compensated quantising error whereas 42 defines a DCT-transformed and movement compensated feedback signal. A re-coded video signal 43 demonstrates lower bit rate than in point 33. In Fig 4, 44 is side information.

In Fig 5, a modifying unit for DCT-components is utilised between the decoder and the coder. In parallel with the unit, a supervising unit is provided for further information. This embodiment operates without loop, compare Fig 4. A variable length decoder VLD is defined by 45 and a modified DCT-unit is disclosed by 46. Also in this case DCT means Discrete Cosines Transform. This component is essential for the present invention since certain selected quantised DCT-components obtain a changed value in a predetermined criteria. 47 defines a variable length coder VLC. A coded video signal 48 is introduced to the variable length decoder 45, which transfers quantised DCT- components to the component 46. Quantised modified DCT- components are disclosed by 50. 51 defines a re-coded video signal with lower bit rate than said coded video signal 48. 52 refers to side information.

Fig 6 discloses the case without loop and with re-quantisation. A variable length decoder VLD is defined by 53 and a variable length coder VLC by 57. An inverse quantiser 1 (IQ₂) is disclosed by 54 and a quantiser 2 (Q₂) is defined by 55. Also in this case, a core component as defined above is used, which core component has the reference sign 56 and refers to a modification of DCT-components. The invention operates in this case with certain selected quantised DCT-components which obtain a changed value according to a predetermined criteria. 58 refers to a coded video signal and 59 to quantised DCT- components. Inverse quantised DCT-components are defined by 60 and re-quantised DCT-components are defined by 61 . 62 refers to re-quantised and modified DCT-components and 63 defines a re-coded video signal which demonstrates a bit rate that is lower than the bit rate of the coded video signal 48. A side information is defined by 64.

Fig 7 intends to disclose the case with image conversion with loop and re-quantisation. The construction corresponds to the embodiment of Fig 4 except for the fact that the quantiser is followed by a unit that modifies DCT. In this case, a variable length decoder VLD is defined by 65, an inverse quantiser 1 (IQ-i) by 66, and a quantiser 2 (Q₂) by 67. Modification of DCT- components takes place in 68, and similarly to the embodiment above, is a core component in the present invention since certain selected quantised DCT-components obtain a changed value according to a predetermined criterion. An inverse quantiser 2 (IQ₂) is disclosed by 69, an Inverse Discrete Cosines Transform (IDCT) by 70, a movement compensator MC using part of the side information by 71 , a Discrete Cosines Transform (DCT) by 72 and a variable length coder (VLC) by 73. A coded video signal has the reference sign 74, quantised DCT- components are defined by 75, inverse quantised DCT- components are defined by 76 and inverse quantised DCT- components are defined by 77 and provided to compensate for quantising errors in previous images. Re-quantised DCT- components have the reference sign 78 and re-quantised and modified DCT-components have the reference sign 79. Modified and inverse quantised DCT-components are defined by 80, and a quantising error of DCT-components is defined by 81 . An inverse transformed quantising error has the reference sign 82, and a movement compensated and inverse transformed quantising error has the reference sign 83. A feedback, movement compensated and DCT-transformed quantising error has been defined by 84. 85 refers to a re-coded video signal with lower bit rate than the one in point 74. A side information is in this case defined by 86.

In the embodiment according to Fig 8 a Discrete Cosines Transform DCT is defined by 87, a quantiser Q by 88, an inverse quantiser IQ by 89, an Inverse Discrete Cosines Transform IDCT by 90, a movement compensator MC using part of the side information by 91 , a decision module ME including movement estimation by 92, and a variable length coder VLC by 93. An uncompress video signal is defined by 94, a prediction error by 95, DCT-components of transformed prediction error by 96, quantised DCT-components by 97, inverse quantised DCT- components by 98, inverse. transformed DCT-components by 99, reconstructed video by 100, prediction, movement compensated reconstructed video by 101 , side information by 102 and coded video by 103.

In Fig 9, a Discrete Cosines Transform DCT is defined by 104, a quantiser Q by 105, and modifying of DCT-components by 106. The latter component is a core component in the present invention, in which certain selected quantised DCT-components obtain a changed value according to a predetermined criterion. By 107 an inverse quantiser is defined, by 108 an inverse Discrete Cosines Transform IDCT, by 109 a movement compensator MC using part of the side information, by 1 10 a decision module ME including movement estimation, by 1 1 1 a variable length coder VLC, by 1 12 a video signal, by 1 13 a prediction error, by 1 14 DCT-components of transformed prediction error, by 1 15 quantised DCT-components by 1 16 modified DCT-components, by 1 17 inverse quantised and modified DCT-components, by 1 18 inverse transformed DCT- components, by 1 19 reconstructed video, by 120 prediction and movement compensated reconstructed video, by 121 side information, and by 122 coded video. The method and the arrangement to change the value of quantised transform components, e. g. lev, is useful also in a coder, i. e. a device where uncompressed, analogue or digital, video is converted to compressed form. A coder according to known technique is illustrated in Fig 8, and a coder supplemented according to the present invention is illustrated in Fig 9. The unit 106 in Fig 9 is thus characterised in that the value of lev is changed to a new value if this involves a saving of at least a predetermined value. 1 12 may also be regarded to symbolise receiving equipment, which receives an uncompressed video signal, and 122 may be regarded to symbolise transmitting equipment, which transmits a compressed video signal.

In the case according to Fig 10 with recording on digital storing media by the use of the present invention, a parabolic antenna 123 is included. In addition, a "Set-top-box" for digital TV- receivers 124, a television apparatus 125, a rate converter 126 according to the present invention, and a digital storing medium 127, for instance a hard disc, are used. 128 illustrates a signal from the parabolic antenna, 129 a decoded video signal, 130 a coded video signal and 131 a converted video signal with lower bandwidth than the coded video signal 130.

Fig 1 1 discloses an example of relay transmission to a transmission network with lower bandwidth. In this case, a parabolic antenna 132, a module 133 for extracting the coded video signal, a rate converter 134 according to the present invention, and a transmission network 135, e. g. Internet, are used. 136 constitutes the signal from the parabolic antenna, 137 a coded video signal, and 138 a converted video signal with lower bandwidth than the coded video signal 137.

In the example according to Fig 12, multiplexing of several video channels is discloses. In this case, 139 constitutes a parabolic antenna, 140 a module for extracting a coded video signal 150, 141 a rate converter according to the present invention, 142 an antenna for ground transmitted digital-TV, 143 a module for extracting a coded video signal 153, 144 a rate converter according to the present invention, 145 a TV-camera, 146 a video coder 147 a rate converter according to the present invention, and 148 a multiplexor. 149 constitutes the signal from the parabolic antenna, 150 a coded video signal, 151 a converted video signal with lower bandwidth than the coded video signal 150, 152 a signal from the TV-antenna, 153 a coded video signal, 154 a converted video signal with lower bandwidth than the last mentioned coded video signal 153, 155 a video signal, 156 a coded video signal, 157 a converted video signal with lower bandwidth than the video signal 156, and the coded and bite rate converted video signals multiplexed to a bit stream 158.

The invention may thus in accordance with the embodiments above use the fact that certain information from an incoming bit stream is reused in the outgoing bit stream. No loop is required, for instance in the video case according to Fig 5, although a loop as disclosed may be used in certain embodiments. The value of "lev" in certain selected DCT-components is changed without the need of re-quantisation.

As much as possible of the incoming bit stream is relay transmitted without being changed. This is valid for SEQ-, GOP-, PIC- and SLI-headers, macro-block type (intra, skipped, MC- type) and movement vectors. The only thing which is modified in an embodiment is DCT-components, compare for instance Fig 5, also vbv-delay in the PIC-header is changed.

In certain cases, for instance critical cases, it may be appropriate to modify other information in the bit stream than the transform components. Such a case may arise if the incoming bit stream contains errors, for instance due to poor receiving conditions. An erroneous bit stream may contain unallowable values of the side information, and it may then be appropriate to assign allowable values to the erroneous parameters.

In for instance Fig 4, a feedback loop is used for avoiding error propagation from image to image. Such a loop may be avoided if intra coding is frequently used in the incoming (and thus outgoing) bit stream. Within the scope of the invention, it is also possible to use such a loop in combination with the proposed embodiments of the present invention.

The transform components do not need to be re-quantised but the same value of "quant" may be transmitted for each macro- block. It is however possible to combine the present invention with re-quantisation. Instead, the values "lev" of the transform components are changed individually according to a particular criterion.

The invention may operate with various criteria for changing of "lev". An individual "lev" may be changed by one unit, e.g. from the value 1 (one) to 0 (zero), if this involves a bit saving of at least ki bits. Two or several adjacent "lev" may be changed by one unit each, e. g. from the value 1 (one) to 0 (zero), if this involves a bit-saving of at least k₂ bits, "lev" may also be changed from e. g. 2 to 0 or from 2 to 1 if this involves a bit- saving by at least k₃ and k₄ bits, respectively. An individual "lev" may be changed from the value 0 (zero) to 1 (one) if this involves a bit saving of at least k₅ bits. The criteria may be selected at use in different cases. The selection of criteria for changing the values of the transform components determines the bit saving and the image quality that are achieved. This may be effectuated by or in a module called buffer control.

The invention is not limited to the embodiments disclosed above by way of example, but may be subjected to modifications within the scope of the following claims and the inventive thought.

List of reference signs

1 . Decoder.

2. Coder. 3. Coded video signal.

4. Decoded video signal.

5. Coded video signal with lower bit rate than in point 1

6. Decoder.

7. Coder. 8. Coded video signal

9. Decoded video signal

10. Coded video signal with lower bit rate than in point 1 .

1 1 . Side information.

12. VLD - Variable Length Decoder 13. IQi - Inverse Quantiser 1

14. IDCT- Inverse Discrete Cosinus Transform

15. DCT - Discrete Cosinus Transform

16. Q₂ - Quantiser 2

17. VLC - Variable Length Coder. 18. Coded still image.

19. Quantised DCT-components.

20. Inverse quantised DCT-components.

21 . Decoded still image.

22. Unquantised DCT-components. 23. Re-quantised DCT-components.

24. Recoded still image, with fewer bits than in point 8.

25. VLD - Variable Length Decoder

26. IQi - Inverse Quantiser 1

27. Q₂ - Quantiser 2 28. IQ₂ - Inverse Quantiser 2

29. IDCT - Inverse Discrete Cosinus Transform

30. MC - Movemnet Compensator, uses parts of the side information.

31 . DCT - Discrete Cosinus Transform 32. VLC - Variable Length Coder 33. Coded video signal.

34. Quantised DCT-components.

35. Inverse quantised DCT-components.

36. Inverse quantised DCT-components compensated for quantising errors in previous images.

37. Re-quantised DCT-components.

38. Inverse ac requantised DCT-components.

39. Quanising errors of DCT-components.

40. Fed back quantising error. 41 . Fed back and movement compensated quantising error.

42. Fed back, movement compensated and DCT-transformed quantising error.

43. Recoded video signal, with lower bit rate than in point 15.

44. Sidoinformation. 45. VLD - Variable Length Decoder

46. mod DCT - the core component in the present invention in which certain selected quantised DCT- components receive a changed value according to a predetermined criteria 47. VLC - Variable Length Coder.

48. Coded video signal.

49. Quantised DCT-components.

50. Quantised and modified DCT-components.

51 . Recoded video signal, with lower bit rate than in point 48. 52. Side information.

53. VLD - Variable Length Decoder

54. VLC - Variable Length Coder

55. IQ₂ - Inverse Quantiser 1

56. Q₂ - Quantiser 2 57. mod DCT - the core component in the present invention in which certain selected quantised DCT- components receive a changed value according to a predetermined criteria 58. Coded video signal. 59. Quantised DCT-components.

60. Inverse quantised DEC-components.

61 . Re-quantised DCT-components.

62. Re-quantised and modified DCT-components.

63. Recoded video signal, with lower bit rate than in point 32. 64. Side information.

65. VLD - Variable Length Decoder

66. \Qι - Inverse Quantiser 1

67. Q₂ - Quantiser 2 68. mod DCT - the core component in the present invention in which certain selected quantised DCT- components receive a changed value according to a predetermined criteria 69. IQ₂ - Inverse Quantiser 2

70. IDCT - Inverse Discrete Cosinus Transform

71 . MC - Movement compensater, uses parts of the side information.

72. DCT - Discrete Cosinus Transform 73. VLC - Variable Length Coder

74. Coded video signal.

75. Quantised DCT-components.

76. Inverse quantised DCT-components.

77. Inverse quantised DCT-components compensated for quantising errors in previous images.

78. Re-quantised DCT-components.

79. Re-quantised and modified DCT-components.

80. Modified and inverse quantised DCT-components.

81. Quantising errors of DCT-components. 82. Inverse transformed quantising error.

83. Movement compensated and inverse transformed quantising error.

84. Fed back, movement compensated and DCT-transformed quantising error. 85. Recoded video signal, with lower bit rate than in point 39.

86. Side information.

87. DCT - Discrete Cosinus Transform

88. Q - Quantiser

89. IQ - Inverse Quantiser 90. IDCT- Inverse Discrete Cosinus Transform

91 . MC - Movemnet compensator, uses parts of the side information.

92. ME - Decsison module including movement estimation.

93. VLC - Variable Length Coder. 94. Video signal.

95. Prediction error.

96. DCT-components of transformed prediction error.

97. Quantised DCT-components.

98. Inverse quantised DCT-components. 99. Inverse transformed DCT-components.

100. Reconstructed video.

101 . Prediction, movement compensated reconstructed video.

102. Side information. 103. Coded video.

104. DCT - Discrete Cosinus Transform

105. Q -Quantiser

106. mod DCT - the core component in the present invention in which certain selected quantised DCT- components receive a changed value according to a predetermined criteria

107. IQ - Inverse Quantiser

108. IDCT- Inverse Discrete Cosinus Transform 109. MC - Movement compensator, uses parts of the side information.

1 10. ME - Decision module including movement estimation.

1 1 1 . VLC - Variable Length Coder.

1 12. Video signal. 1 13. Prediction error.

1 14. DCT-components of transformed prediction error.

1 15. Quantised DCT-components.

1 16. Modified DCT-components.

1 17. Inverse quantised and modified DCT-components. 1 18. Inverse transformed DCT-components.

1 19. Reconstructed video.

120. Prediction, movement compensated reconstructed video.

121. Side information.

122. Coded video. 123. Parabolic antenna.

124. Set-top-box, digital-TV-receiver.

125. Television device

126. Rate converter according to the present invention.

127. Digital storing medium, e.g. a hard disc. 128. Signal from a parabolic antenna.

129. Decoded video signal.

130. Coded video signal.

131 . Converted video signal, with lower bandwidth than 130.

132. Parabolic antenna 133. Module for extracting of ceded video signal.

134. Rate converter according to the present invention.

135. Transmission network, e.g. internet.

136. Signal from parabolic antenna.

137. Coded video signal. 138. Converted video signal, with lower bandwidth than 137.

139. Parabolic antenna.

140. Module for extracting of coded video signal 150.

141 . Rate converter according to the present invention. 142. Antenna for ground transmitted digital-TV.

143. Module for extracting of coded video signal 150.

144. Rate converter according to the present invention.

145. TV-camera. 146. Video coder.

147. Rate converter according to the present invention.

148. Multiplexer,

149. Signal from the parabolic antenna.

150. Coded video signal. 151 . Converted video signal, with lower bandwidth than 150.

152. Signal from TV-antenna.

153. Coded video signal.

154. Converted video signal, with lower bandwidth than 153.

155. Video signal. 156. Coded video signal.

157. Converted video signal, with lower bandwidth than156.

158 The coded and bit rate converted video signals multiplexed to a bit stream

Claims

1. A device for providing change of a video signal, which is subjected to a transformation utilising quantised transform components and operates with supporting information, characterised in that the device comprises a unit which is arranged to modify selected quantised DCT-components according to a predetermined criterion and to modify substantially solely the transform components in one of the following two cases: a) where the device forms a unit converting or reducing the video signal, wherein the unit includes a decoder, which receives the video signal, and a coder, which effectuates a change in the form of a conversion or reduction of the video signal received, wherein said modification results in at least substantially no influence on the supporting information. b) where the device forms a coder receiving the video signal, wherein the coder is arranged to provide by itself a change in the form of a compression of the video signal received and to generate the supporting information substantially independently of the compression.

2. A device according to claim 1 a, characterised in that also the video buffer verification delay in the PIC-header is subjected to a change function.

3. A device according to claim 1 a or 2, characterised in that macro-block type and movement vectors are subjected to change functions in case these have unallowable values.

4. A device according to claim 1 a, 2 or 3, characterised in that the device operates in a circuit with or without a feedback circuit arranged to avoid error propagation from image to image.

5. A device according to claim 1a, 2, 3 or 4, characterised in that the device during the modification of the transform component or transform components operates with or without re- quantisation.

6. A device according to any one of claims 1 - 5, characterised in that the device is arranged to change the value of one or several transform components, e.g. lev, according to a selected criteria, e.g. a criteria based on a large bit-saving with a simultaneous small reduction, or substantial maintaining, of the image quality.

7. A device according to any one of claims 1 - 6, characterised in that the device is arranged to operate with one or several of the following alternatives: a) lack of modification of transform component in certain selected DCT-blocks, b) modification of an individual transform component in certain selected DCT-blocks, c) modification of a number of transform components in certain selected DCT-blocks, d) modification of all transform components in certain selected DCT-blocks, e) modification of all transform components in certain selected DCT-blocks.

8. A device according to any one of claims 1 - 7, characterised in that an individual transform component is changeable by one unit, if this involves a bit-saving by a predetermined minimum number of bits.

9. A device according to any one of claims 1 - 8, characterised in that an individual transform component is changeable by more than one unit, if this involves a bit-saving by a predetermined minimum number of bits.

10. A device according to any one of claims 1 - 9, characterised in that several transform components together are changeable with one or several units if this involves a total bit saving by a predetermined minimum number of bits.

1 1. A device according to any one of claims 1 - 10, characterised in that a coder includes a unit (mod DCT 106) for modifying quantised DCT-components.

12. A device according to claim 1 1 , characterised in that the unit (mod DCT 106) receives a signal (1 15) from a quantiser (Q-, 105) and with its output is connected to a variable length-coder (VLC 1 1 1 ) and a feedback path.

13. A system including one or several devices for converting, reducing or compressing a video signal, which is subjected to a transformation utilising quantised transform components with code words and supplied with or includes supporting information, characterised in that the respective device for providing said influence on the video signal either includes a decoder, which receives the actual signal to be influenced, and a coder, which provides the video signal in a converted or reduced form, or in itself forms a coder, which compresses the video signal received, that the respective device comprises a unit which is arranged to modify selected quantised DCT-components according to a predetermined criterion, to modify substantially solely the transform components, and that the supporting information in such a way is substantially not influenced, and that the device concerned is included in or connectable to any one of the following functional alternatives: a) recording on digital storing media for providing a memory saving and/or modified image quality, b) relay transmission in or to a transmission network with smaller bandwidth, c) multiplexing of two or several video channels, d) transfer from variable to constant bit rate, e) transfer from constant to variable bit rate, f) transfer from constant to constant lower bit rate, g) transfer from variable to variable lower bit rate, and/or h) compression of the video signal between transmission and receiving equipment or vice versa.