Coding method of image information

A coding method of a binary Markov information source comprises the steps of providing a range on a number line from 0 to 1 which corresponds to an output symbol sequence from the information source, and performing data compression by binary expressing the position information on the number line corresponding to the output symbol sequence. The present method further includes the steps of providing a normalization number line to keep a desired calculation accuracy by expanding a range of the number line which includes a mapping range, by means of a multiple of a power of 2, when the mapping range becomes below 0.5 of the range of the number line; allocating a predetermined mapping range on the normalization number line for less probable symbols LPS proportional to its normal occurrence probability; allocating the remaining mapping range on the normalization number line for more probable symbols MPS; and reassigning the predetermined mapping range to the remaining mapping range the half of a portion where the allocated remaining range is less than 0.5, when the allocated remaining range becomes below 0.5.

Skip to:  ·  Claims  ·  References Cited  · Patent History  ·  Patent History

Claims

1. A method for coding information from a binary Markov information source by binary coding an output symbol sequence from said information source comprising less probable symbols (LPS) and more probable symbols (MPS), each having an occurrence probability, on a normalization number line, said method comprising the steps of;

a) storing in a memory storage device a normalization number line having a range from 0 to 1 which corresponds to said output symbol sequence,
b) keeping a desired calculation accuracy by expanding a range of the normalization number line which includes a mapping range by means of a multiple of a power of 2 when the mapping range becomes less than 0.5,
c) allocating a portion of said normalization number line as a predetermined mapping interval for said LPSs, said portion being proportional to the occurrence probability of said LPSs,
d) allocating the remaining portion of said number line as a mapping interval for said MPSs,
e) reassigning half of the LPS mapping interval above 0.5 to said MPS mapping interval when the LPS mapping range exceeds 0.5, and
f) repeating steps b, c, d and e.

2. A coding method as set forth in claim 1 whereas said LPS mapping interval is a power of 1/2 of the range of said number line.

4. An apparatus for coding information from a binary Markov information source by binary coding an output symbol sequence comprising less probable symbols (LPSs) and more probable symbols (MPS) from said information source on a normalization number line, said LPSs and MPSs each having an occurrence probability, said apparatus comprising:

memory storage means for storing a normalization number line having a range from 0 to 1 which corresponds to said output symbol sequence,
means for keeping desired calculation accuracy by expanding a range on said normalization number line, which includes a mapping range, by a multiple power of 2 when the mapping range becomes less than 0.5,
means for allocating a portion of said normalization number line as a predetermined mapping interval for said LPSs, said portion being proportional to the occurrence probability of said LPSs,
means for allocating the remaining portion of said normalization number line as a mapping interval for said MPSs,
means for reassigning half of the LPS mapping interval above 0.5 to said MPS mapping interval when said LPS mapping interval exceeds 0.5.

5. An apparatus as set forth in claim 4 wherein said LPS mapping interval is a power of 1/2 of the range of said number line.

7. A method for coding information from a Markov information source by binary coding an output symbol sequence from said information source comprising less probable symbols (LPSs) and more probable symbols (MPSs), each sequence having an occurrence probability on a number line, said method comprising,

(a) storing in memory storage device a number line having a range which corresponds to said output symbol sequence;
(b) allocating a portion of said number line as a predetermined mapping interval for said LPSs, said portion being proportional to the occurrence probability of said LPS;
(c) allocating the remaining portion of said number line as a mapping interval for said MPSs; and
(d) controlling the allocating portion of said number line as a mapping interval for said LPSs by assigning a predetermined portion of the mapping interval for said LPSs above a prescribed value of said number line to said mapping interval for said MPSs, so as to maintain said portion proportional to the occurrence probability of said LPSs..Iaddend..Iadd.

8. An apparatus for coding information from a Markov information source by binary coding an output symbol sequence from said information source comprising less probable symbols (LPSs) and more probable symbols (MPSs) from said information source on a number line, said LPSs and MPSs each having an occurrence probability, said apparatus comprising:

memory storage means for storing a number line having a range which corresponds to said output symbol sequence;
means for allocating a portion of said number line as a predetermined mapping interval for said LPSs, said portion being proportional to the occurrence probability of said LPSs;
means for allocating the remaining portion of said number line as a mapping interval for said MPSs; and
control means for controlling the allocating portion of said number line as a mapping interval for said LPSs by assigning a predetermined portion of the mapping interval for said LPSs above a prescribed value of said number line to said mapping interval for said MPSs, so as to maintain said portion proportional to the occurrence probability of said LPSs..Iaddend..Iadd.9. A method for coding information from a Markov information source by binary coding an output symbol sequence from said information source comprising less probable symbols (LPSs) and more probable symbols (MPSs) each having an occurrence probability on a number line, said method comprising,
(a) storing in memory storage device a number line having a range which corresponds to said output symbol sequence;
(b) allocating a a portion of said number line as a predetermined mapping interval for said LPSs, said portion being proportional to the occurrence probability of said LPSs;
(c) allocating the remaining portion of said number line as a mapping interval for said MPSs; and
(d) reassigning half of the LPSs mapping interval above a prescribed value to said MPSs mapping interval when the LPSs mapping range exceeds the prescribed value, and
(e) repeating steps b, c, and d..Iaddend..Iadd.10. An apparatus for coding information from a Markov information source by binary coding an output symbol sequence from said information source comprising less probable symbols (LPSs) and more probable symbols (MPSs) from said information source on a number line, said LPSs and MPSs each having an occurrence probability, said apparatus comprising:
memory storage means for storing a number line having a range which corresponds to said output symbol sequence;
means for allocating a a portion of said number line as a predetermined mapping interval for said LPSs, said portion being proportional to the occurrence probability of said LPSs;
means for allocating the remaining portion of said number line as a mapping interval for said MPSs; and
means for reassigning half of the LPSs mapping interval above a prescribed value to said MPSs mapping interval when said LPSs mapping range exceeds

the prescribed value..Iaddend..Iadd.11. A decoding method for a Markov information source coded by binary coding comprising the steps of:

associating more probable symbols (symbols of a higher occurrence probability) and less probable symbols (symbols of a lower occurrence probability) to predetermined ranges on a number line on the basis of a range on a number line for a preceding symbols;
outputting a decoding signal according to a result of correspondence between the ranges and an inputted codeword;
comparing the range on the number line of more probable symbols with the range on the number line of less probable symbols; and
adjusting the range on the number line of less probable symbols and the range on the number line of more probable symbols by assigning predetermined portion of the range for said less probable symbols above a prescribed value of said number line to said range for said more probable symbols so that the range on the number line of less probable symbols does

not exceed that of the more probable symbols..Iaddend..Iadd.12. A decoding method for a Markov information source coded by binary coding comprising the steps of:

associating more probable symbols (symbols of a higher occurrence probability) and less probable symbols (symbols of a lower occurrence probability) to predetermined ranges on a number line on the basis of a range on a number line for a preceding symbols;
outputting a decoding signal according to a result of correspondence between the ranges and an inputted codeword;
comparing a range on the number line of more probable symbols with a fixed value; and
adjusting the range on the number line of more probable symbols and the range on the number line of less probable symbols so that when a range of more probable symbols is below the fixed value on a number line, half of a value below the fixed value of a range more probable symbols is moved from the range of less probable symbols to that of more probable symbols.

.Iaddend..Iadd.13. A coding method for a Markov information source by binary coding comprising the steps of:

associating more probable symbols (symbols of a higher occurrence probability) and less probable symbols (symbols of a lower occurrence probability) to predetermined ranges on a number line on the basis of a range on a number line for a preceding symbols;
coding a signal according to a result of correspondence between the ranges to generate a codeword;
comparing the range on the number line of more probable symbols with the range on the number line of less probable symbols; and
adjusting the range on the number line of less probable symbols and the range on the number line of more probable symbols by assigning a predetermined portion of the range for said less probable symbols above a prescribed value of said number line to said range for said more probable symbols so that the range on the number line of less probable symbols does

not exceed that of the more probable symbols..Iaddend..Iadd.14. A coding method for a Markov information source by binary coding comprising the steps of:

associating more probable symbols (symbols of higher occurrence probability) and less probable symbols (symbols of a lower occurrence probability) to predetermined ranges on a number line on the basis of a range on a number line or a preceding symbols;
coding a signal according to a result of correspondence between the ranges to generate a codeword;
comparing a range on the number line of more probable symbols with a fixed value; and
adjusting the range on the number line of more probable symbols and the range on the number line of less probable symbols so that when a range of more probable symbols is below the fixed value on a number line, half of a value below the fixed value of a range of more probable symbols is moved from the range of less probable symbols to that of more probable symbols..Iaddend.
Referenced Cited
U.S. Patent Documents
4028731 June 7, 1977 Arps et al.
4070694 January 24, 1978 Sakamoto et al.
4099257 July 4, 1978 Arnold et al.
4177456 December 4, 1979 Fukinuki et al.
4191974 March 4, 1980 Ono et al.
4286256 August 25, 1981 Langdon, Jr. et al.
4355306 October 19, 1982 Mitchell
4905297 February 27, 1990 Langdon, Jr. et al.
4933883 June 12, 1990 Pennebaker et al.
Other references
  • Pennebaker et al., An Overview of the Basic Priciples of the Q-Coder Adaptive Binary Arithmetic Coder, IBM Journal of Research and Development, vol. 32, No. 6, Nov. 1988, pp. 717-726. K. S. Fu et al., Robotics: Control, Sensing, Vision, and Intelligence, McGraw-Hill Book Company, New York, copyright 1987, pp. 342-351.
Patent History
Patent number: RE35781
Type: Grant
Filed: Nov 7, 1995
Date of Patent: May 5, 1998
Assignee: Mitsubishi Denki Kabushiki Kaisha (Tokyo)
Inventors: Fumitaka Ono (Kamakura), Shigenori Kino (Kamakura), Masayuki Yoshida (Kamakura), Tomohiro Kimura (Kamakura)
Primary Examiner: Howard L. Williams
Law Firm: Wolf, Greenfield & Sacks, P.C.
Application Number: 8/553,235
Classifications
Current U.S. Class: Adaptive Coding (341/51); To Or From Code Based On Probability (341/107); 358/2612
International Classification: H03M 734;