Process for the transmission of images

• US 5,771,102 A
• Filed: 12/22/1995
• Issued: 06/23/1998
• Est. Priority Date: 03/11/1994
• Status: Expired due to Term

• Current Assignee: IPCom GmbH & Co KG
• Original Assignee: Robert Bosch GmbH
• Inventors: Lappe, Dirk, Vogt, Oliver, Pelz, Rodolfo Mann, Vogt, Wilhelm, Nitsche, Gunnar, Mundt, Frieder
• Primary Examiner(s): Vu, Kim Yen
• Assistant Examiner(s): NGUYEN, ANH-VINH THI
• Application Number:
  US08/553,467
• Time in Patent Office: 914 Days
• Field of Search: 358/433, 358/426, 358/432, 358/261.4, 358/467, 358/261.1, 358/261.2, 358/431, 358/405, 358/412, 341/50, 341/52, 341/55, 341/56, 341/59, 341/67, 341/123, 371/37.1, 371/47.1
• US Class Current: 358/426.02
• CPC Class Codes: H04L 2001/0098 Unequal error protection
• CPC Class Codes: H04N 19/00 Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
• CPC Class Codes: H04N 19/61 in combination with predictive coding
• CPC Class Codes: H04N 19/70 characterised by syntax aspects related to video coding, e.g. related to compression standards
• CPC Class Codes: H04N 19/89 involving methods or arrangements for detection of transmission errors at the decoder
• CPC Class Codes: H04N 7/52 Systems for transmission of a pulse code modulated video signal with one or more other pulse code modulated signals, e.g. an audio signal or a synchronizing signal assembling of a multiplex stream by combining a video stream with other content or additional data, remultiplexing of multiplex streams, insertion of stuffing bits into the multiplex stream, assembling of a packetised elementary stream at server side H04N21/236; disassembling of a multiplex stream, remultiplexing of multiplex streams, extraction or processing of Service Information, disassembling of packetised elementary stream at client side H04N21/434

Abstract

The process for transmission of images enables various classes of source-coded image data to be channel-coded and transmitted at various class code rates. For this purpose, source-coded image data to be transmitted are divided into classes of different sensitivity with respect to transmission errors and the classes are provided with different error protection and are channel-coded for transmission. Information words are provided at the beginning of each class containing information regarding the class type, code rate, number of data bits belonging to a particular class and regarding the error protection employed. In addition, a synchronizing word is added to the image data for each image. In the receiver the first information words are channel-decoded according to a fixed code rate and the information stored in the first information words including code rate is used for the channel-decoding of the received coded image data. The image data are subsequently subjected to an image source decoding and read out.

Claims

1. A process for transmission of data comprising the steps of:

a) generating said data by at least one data source;

b) inserting said data in a data frame at a channel coder;

c) dividing said data into a plurality of classes of variable length;

d) preparing an information word of fixed length for each of said classes, said information word containing information regarding class type, number of data bits therein and class code rate, wherein said number of said data bits and said class code rate can be different for different ones of said classes;

e) channel-coding said data for each of said classes by first reading said information word with said channel coder to obtain said class code rate and then channel-coding at said class code rate obtained from the reading to obtain channel-coded data; and

f) channel-coding said information word for each of said classes at a fixed code rate to obtain a channel-coded information word for each of said classes.

2. The process according to claim 1, further comprising storing information regarding data generation and/or data source coding as data of a class in the data frame.

3. The process according to claim 1, wherein each of said classes contains image data, audio data, facsimile data and/or data files.

4. The process according to claim 1, further comprising subjecting the channel-coded data of each of said classes to an interleaving after said channel-coding, wherein during the interleaving the channel-coded data are read into diagonal positions in a first memory matrix and subsequently read out line by line from the first memory matrix.

5. The process according to claim 4, wherein the data frame contains a fixed n umber of said data bits corresponding to storage locations for a whole-number quantity o f said diagonal positions in the first memory matrix.

6. The process according to claim 1, further comprising synchronizing the transmission of said channel-coded data by a pseudo-noise sequence with autocorrelation characteristics.

7. The process according to claim 6, wherein said synchronizing is performed by a synchronizing method including forming a synchronizing word consisting of a pseudo-noise word composed of two Williard code words or two Barker code words.

8. The process according to claim 1, further comprising channel-decoding the respective information words at a fixed decoding rate to determine said class code rates and said numbers of data bits for the respective classes and subsequently channel-decoding said channel-coded data in the respective classes with the respective class code rates determined from said information words.

9. The process according to claim 8, wherein said channel decoding is performed by a Viterbi decoder operating by the maximum likelihood sequence estimation method.

10. The process according to claim 9, wherein said Viterbi decoder is in a known state after a last data bit of the channel-coded information word is read in for each of said classes, said channel-coded information word is evaluated by the Viterbi decoder without reading in additional data bits of the data frame, since bits are read into the Viterbi decoder which do not alter the state of the Viterbi decoder.

11. The process according to claim 1, wherein said information word for each of said classes is terminated by a fixed number of zero bits (15) or one bits, said channel coder has a memory depth and said fixed number corresponds to said memory depth.

12. A process for transmission of channel-coded source-coded image data over a data channel, said process comprising the steps of:

a) dividing the image into macroblocks;

b) source-coding each of said macroblocks to obtain source-coded image data;

c) dividing said source-coded image data for said macroblocks into respective pluralities of classes of said image data, said classes having respective different image reconstruction sensitivities to transmission errors;

d) providing each of said classes with a first information word (9) having a fixed structure and containing information regarding class type, channel-coding, class coding rate and an amount of image data present herein, wherein said amount of image data and said class code rate are variable among said classes;

e) adding a fixed number of code words (8) to said source-coded image data prior to channel coding, said code words (8) containing information regarding the type of said source coding;

f) providing a second information word (7) together with said code words (8), said second information word having a fixed pattern and containing information relating to code word type, channel-coding type and number of data bits in said code words (8);

g) channel-coding and transmitting for each of said classes said first information word (9) and said second information word (7) with a predetermined channel coding;

h) channel-coding and transmitting said code words (8) and said source-coded image data (10) of each of said classes with a class-determined channel coding at said class coding rate to form channel-coded source-coded image data; and

i) adding a synchronizing word to the channel-coded source-coded image data.

13. The process according to claim 12, wherein said synchronizing word consists of a pseudo-noise data word composed of two Williard code words or two Barker code words.

14. The process according to claims 12, wherein the macroblocks are transmitted sequentially one after the other.

15. The process according to claim 12, wherein the first and second information word (9,7) each have a class number (12) with a length of four bits, a code pointer (13) with a length of five bits, and a length number (14) with a length of fourteen bits, the class number (12) indicates the class of the associated data, the code pointer (13) indicates the code rate used for the channel-coding, and the length number (14) indicates the number of code word data bits or of class data bits.

16. The process according to claim 12, wherein first and second information word (9, 7) are each terminated by a fixed number of zero bits (15) or one bits and said fixed number corresponds to a memory depth of a channel coder used for the channel-coding.

17. The process according to claim 12, wherein the code words (8) comprise a first and a second code word (16, 17), the first code word (16) has a length of three bits and comprises a first bit, a second bit and a third bit and the first bit indicates image data formation speed, the second bit indicates whether or not one of two possible source coding methods was used for said source-coding, and the third bit indicates whether or not information words are present, and the second code word (17) has a length of five bits and indicates a step value for quantization used to determine DCT coefficients.

18. The process according to claim 17, wherein the code words (8) comprise a third and a fourth code word (18, 19), the third code word (18) has a length of one bit and indicates whether or not additional information about the source coding is stored in the fourth code word (19) which has a length of eight bits.

19. The process according to claim 12, wherein the source-coded image data comprises movement vectors (MV), macroblock positions (P), DCT coefficients (DCT), and quantizing step (QU);

and further comprising arranging said movement vectors (MV), said macroblock positions (P), said DCT coefficients (DCT) and said quantizing step (QU) in respective different ones of said classes.

20. The process according to claim 19, wherein the movement vectors (MV), as one portion of said data, and the DCT coefficients (DCT), as another portion of said data, contain transform coefficients for variable two-dimensional length coding and two bits for identifying a macroblock end and other portions of said data contain information regarding differential quantization (DQUANT) and/or regarding luminance blocks contained in said macroblocks and coded transform coefficients (CBPY) and/or information regarding the luminance blocks of said macroblocks which contain coded transform coefficients, prediction modes used for the image or the macroblocks, and quantization methods employed (CBPCM) and/or information regarding whether or not coding is carried out (COD).

21. The process according to claim 15, wherein the code pointer (13) designates an unspecified code for future classes.

22. The process according to claim 21, wherein the unspecified code is a rate-compatible punctured convolutional code.

23. The process according to claim 15, wherein when said length number (14) indicates that a provided amount of image data (10) is greater than the designated number of said class data bits, the length number is then set to a value equal to (2¹⁴ -1) and said class having said length number reset to said value is followed by another class for which said first information word (9) contains additional data from said class.

24. The process according to claim 12, further comprising feeding the channel-coded source-coded image data into a first memory matrix (23) in a convolutional interleave, and wherein the first memory matrix (23) has a fixed width and is rewritten cyclically by means of the convolutional interleave, the channel-coded source-coded image data are read in bit by bit in diagonal positions in the first memory matrix from the upper left side to the bottom right side and successive ones of the diagonal positions are offset horizontally by one storage location in each instance, and for the purpose of transmission said channel-coded source-coded image data are read out of successive lines of said first memory matrix bit by bit in the sequence in which the diagonal positions were offset relative to one another when read in.

25. The process according to claim 24, wherein a total number of said class data bits to be transmitted for an image correspond to a fixed quantity which corresponds to a multiple of the number of bits which can be read into a diagonal of the first memory matrix (23).

26. The process according to claim 12, wherein the first or second information words include coded information instead of said information regarding said code rate and further comprising decoding said coded information in said channel-decoding to indicate that said information word including said coded information should not be used and a subsequent one of said information words should be used for said channel-decoding.

27. The process according to claim 12, further comprising feeding the channel-coded source-coded image data to a channel decoder, achieving time synchronization from the synchronizing word (6) by correlation comparison, decoding the first information word (9) for each of said classes to obtain said information regarding the class type, the channel-coding, the class coding rate and the amount of image data present therein for each of said classes as well as error protection information;

after decoding the first information words (9) channel-decoding the channel-coded source-coded image data associated with the respective first information words (9) using said information in the first information words (9) to obtain the source-coded image data;

then source-decoding the source-coded image data to obtain decoded reproducible image data and subsequently reproducing the image from the decoded reproducible image data.

28. The process according to claim 27, wherein the second information word (7) has a fixed structure, a fixed bit length and a fixed channel coding, and further comprising decoding said second information words (7) to obtain for each of said classes said information regarding said code word type, said channel-coding type and said number of data bits in said code words as well as error protection information;

after the decoding the respective second information words (7) channel-decoding said code words (8) belonging to the respective second information words (7) using the information stored therein and using the code words (8) for said source-decoding of each of said classes.

29. The process according to claim 27, further comprising feeding said channel-coded source-coded data into a second memory matrix (24) of a convolutional deinterleaver, said second memory matrix (24) of the deinterleaver having a predetermined width and being rewritten cyclically, so that said channel-coded source-coded data are read in bit by bit into diagonal positions from the bottom left side to the top right side of the second memory matrix and successive diagonal positions are offset relative to one another horizontally by one storage location, and reading said channel-coded source-coded data out of said second memory matrix bit by bit and line by line and feeding said channel-coded source-coded data so read out to the channel decoder in the sequence in which said data is read out.

30. The process according to claim 27, wherein said channel-decoding is performing with a Viterbi decoder used as a channel decoder, and the Viterbi decoder operates using the maximum likelihood sequence estimation method.

31. The process according to claim 30, wherein the Viterbi decoder has a memory with a fixed memory depth, and further comprising after detecting a zero bit in the first or second information word (9, 7) supplying said memory of the Viterbi decoder with zero bits from a digital serial processor (3) so that additional image data is not processed, and channel-decoding the first and/or second information word (9, 7) and data entities which follow the first or second information word (9, 7).

32. A process for transmitting image data derived from images, said process comprising the steps of:

a) source-coding the image data of an image to obtain source-coded image data;

b) dividing the source-coded image data of the image into a plurality of classes having respective different image reconstruction sensitivities to transmission errors due to transmitting the image data;

c) adding a first information word (9) to each of said classes, said first information word (9) having a fixed pattern and containing information regarding class type, channel-coding, number of bits in the associated image data, sensitivity to transmission errors and class code rate, wherein said number of bits in the associated image data and said class code rate are variable for each of said classes and said class code rate;

d) transmitting said first information word s (9) for said classes with fixed channel coding and said source-coded image data of said classes after a predetermined channel-coding at said class code rate to obtain channel-coded source-coded image data for the transmission; and

e) adding a synchronizing word to the channel-coded source-coded image data and the first information words (9).

33. A process for sending and receiving image data for an image, said process comprising the steps of:

a) dividing the image into macroblocks;

b) source-coding each of said macroblocks to obtain source-coded image data;

c) dividing said source-coded image data of said macroblocks into respective pluralities of classes, said classes having respective different image reconstruction sensitivities to transmission errors;

d) providing each of said classes with a first information word (9) having a fixed structure and containing information regarding class type, channel-coding, class coding rate and an amount of image data present herein, wherein said amount of image data and said class code rate are variable for each of said classes;

e) adding a fixed number of code words (8) to said source-coded image data prior to channel coding, said code words (8) containing information regarding the type of said source coding;

f) providing a second information word (7) together with said code words (8), said second information word having a fixed pattern and containing information relating to code word type, channel-coding type and number of data bits in said code words (8);

g) channel-coding and transmitting said first information words (9) and said second information words (7) of said classes with a predetermined channel coding;

h) channel-coding and transmitting said code words (8) and said source-coded image data (10) of each of said classes with a class-determined channel coding at said class coding rate to form channel-coded source-coded image data;

i) adding a synchronizing word to the channel-coded source-coded image data;

j) receiving the channel-coded source-coded image data and feeding the channel-coded source-coded image data to a channel decoder;

k) time synchronizing using said synchronizing word (6) by correlation comparison;

l) decoding the respective first information words (9) for said classes to obtain said information regarding the class type, the channel-coding, the class coding rate and the amount of image data present therein for each of said classes as well as error protection information;

m) after said decoding of said first information words, channel-decoding the channel-coded source-coded image data associated with the respective first information words (9) using said information in the first information words (9) to obtain the source-coded image data; and

n) then source-decoding the source-coded image data to obtain decoded reproducible image data.

34. A process for transmission of channel-coded source-coded image date over a data channel, said process comprising the steps of:

a) dividing the image into macroblocks;

b) source-coding each of said macroblocks to obtain source-coded image data;

c) dividing said source-coded image data of said macroblocks into respective pluralities of classes, said classes having respective different image reconstruction sensitivities to transmission errors;

d) providing each of said classes with a first information word (9) having a fixed structure and containing information regarding class type, channel-coding, class coding rate and an amount of image data present herein, wherein said amount of image data and said class code rate a re variable for each of said classes;

e) adding a fixed number of code words (8) to said source-coded image data prior to channel coding, said code words (8) containing information regarding the type of said source coding;

f) providing a second information word (7) together with said code words (8), said second information word having a fixed pattern and containing information relating to code word type, channel-coding type and number of data bits in said code words (8);

g) channel-coding and transmitting said first information words (9) and said second information words (7) for said classes with a predetermined channel coding;

h) channel-coding and transmitting said code words (8) and said source-coded image data (10) of each of said classes with a class-determined channel coding at said class coding rate to form channel-coded source-coded image data; and

i) adding a synchronizing word to the channel-coded source-coded image data;

wherein the source-coded image data comprises movement vectors (MV), macroblock positions (P), DCT coefficients (DCT), and quantizing step (QU); and

further comprising arranging said movement vectors (MV), said macroblock positions (P), said DCT coefficients (DCT) and said quantizing step (QU) in respective different ones of said classes.