Method and apparatus fr arithmetic coding
Method and apparatus fr arithmetic coding
 CN 1,675,842 A
 Filed: 09/19/2003
 Published: 09/28/2005
 Est. Priority Date: 09/20/2002
 Status: Active Application
First Claim
1. method of coding data, described method comprises:
 A plurality of incidents in the coding sequence of events produce coded data;
AndUtilize coded data to produce bit stream, comprise zero or a plurality of filling bit added in the bit stream after the coded data that wherein said zero or a plurality of filling bit are operated the relation between the bit number that is used for keeping the piece number of encoding event number, encoding and bit stream in fact.
Chinese PRB Reexamination
Abstract
Methods and apparatuses for performing arithmetic encoding and/or decoding are disclosed. In one embodiment, an arithmetic decoder comprises a sequencer to generate a context identifier for an event of an event sequence, a probability estimator to determine a value for a LPS and a probability estimate for the LPS, and a decoding engine that includes a range register to assign a value to a range for the LPS. The value is based on the probability estimate, a value stored in the range register and the context identifier to a range for the LPS if the context identifier is not equal to an index and the value is not based on the value stored in range register if the context identifier is equal to the index. The decoding engine further determines a value of a binary event based on the value of the range for the LPS and bits from an information sequence.

7 Citations
Method and device for arithmetic encoding or arithmetic decoding  
Patent #
US 10,848,180 B2
Filed 11/07/2019

Current Assignee
N/A

Method and device for arithmetic encoding or arithmetic decoding  
Patent #
US 10,516,414 B2
Filed 04/12/2018

Current Assignee
N/A

Combined blocksymbol error correction  
Patent #
CN 104,508,982 A
Filed 10/31/2012

Current Assignee

Method and device for arithmetic encoding or arithmetic decoding  
Patent #
CN 104,993,833 A
Filed 10/01/2010

Current Assignee

Method and device for arithmetic encoding or arithmetic decoding  
Patent #
CN 105,099,463 A
Filed 10/01/2010

Current Assignee

Method and device for arithmetic encoding or arithmetic decoding  
Patent #
US 9,973,208 B2
Filed 10/27/2015

Current Assignee
N/A

Techniques for contextadaptive binary data arithmetic coding (CABAC) decoding  
Patent #
CN 103,636,206 A
Filed 03/31/2011

Current Assignee

No References
94 Claims

1. method of coding data, described method comprises:

A plurality of incidents in the coding sequence of events produce coded data;
AndUtilize coded data to produce bit stream, comprise zero or a plurality of filling bit added in the bit stream after the coded data that wherein said zero or a plurality of filling bit are operated the relation between the bit number that is used for keeping the piece number of encoding event number, encoding and bit stream in fact.


2. method according to claim 1 is characterized in that:
 described one or more filling bits comprise that the front has one or more byte of paddings of zero or a plurality of alignment bits.

3. method according to claim 2 is characterized in that:
 described one or more byte of paddings are can be by the decoder recognized patterns.

4. method according to claim 1 is characterized in that:
 described arithmetic encoder is used for producing coded data according to sequence of events.

5. method according to claim 4 is characterized in that:
 described bit stream comprises header data.

6. method according to claim 1 is characterized in that:
 described filling bit is added in the bit stream after the coded data, and described coded data comprises the ending of the fragment indication of having encoded.

7. method according to claim 5 is characterized in that also comprising that with image segmentation be described one or more fragment, and wherein each of each fragment includes one or more macro blocks.

8. arithmetic encoder comprises:

The probability Estimation device, each incident that is used to produce sequence of events all has the probability Estimation of particular value, and wherein the response of probability Estimation device produces described probability Estimation at the respective environment information of described each incident;
AndThe coding engine that links to each other with the probability Estimation device, be used to respond each incident and corresponding probability Estimation thereof, produce zero or a plurality of bit of information sequence, wherein coding engine produces zero or a plurality of filling bit append to after the bit that response events produced on the information sequence, and wherein said one or more filling bits operations are used for keeping in fact the amount of the incident of encoding, just at coded data piece number and by the relation of coding engine between the bit number that information sequence produces.


9. encoder according to claim 8 is characterized in that described one or more filling bit comprises that the front has one or more byte of paddings of zero or a plurality of alignment bits.

10. encoder according to claim 9 is characterized in that described one or more byte of padding is can be by the decoder recognized patterns.

11. encoder according to claim 8 is characterized in that described arithmetic encoder is used for producing coded data according to sequence of events.

12. encoder according to claim 8 is characterized in that described bit stream comprises header data.

13. encoder according to claim 8 is characterized in that described coding engine adds described filling bit in the bit stream to after the coded data, described coded data comprises the ending of the fragment indication of having encoded.

14. the manufacture with one or more recordable medias of having stored instruction on it, described instruction is carried out by system, makes system come coded data in the following manner:

A plurality of incidents in the coding sequence of events produce coded data;
AndUtilize coded data to produce bit stream, comprise zero or a plurality of filling bit added to after the coded data that the response events sequence is produced in the bit stream that wherein said zero or a plurality of filling bit are operated the relation between the bit number that is used for keeping the piece number of encoding event number, encoding and bit stream in fact.


15. manufacture according to claim 14 is characterized in that:
 described one or more filling bits comprise that the front has one or more byte of paddings of zero or a plurality of alignment bits.

16. manufacture according to claim 15 is characterized in that:
 described one or more byte of paddings are can be by the decoder recognized patterns.

17. manufacture according to claim 14 is characterized in that:
 described filling bit is added in the bit stream after the coded data, and described coded data comprises the ending of the fragment indication of having encoded.

18. the equipment of a coded data, described equipment comprises:

Device, a plurality of incidents of the sequence of events that is used for encoding produce coded data;
AndDevice, utilize coded data to produce bit stream, comprise zero or a plurality of filling bit added to after the coded data that the response events sequence is produced in the bit stream that wherein said zero or a plurality of filling bit are operated the relation between the bit number that is used for keeping the piece number of encoding event number, encoding and bit stream in fact.


19. an arithmetic decoder comprises
The probability Estimation device, each incident that is used to produce sequence of events all has the probability Estimation of particular value, and wherein the response of probability Estimation device produces described probability Estimation at the respective environment information of the incident of described sequence of events;  And
The Decode engine that links to each other with the probability Estimation device, be used to respond its corresponding probability Estimation and information sequence, produce the incident of sequence of events, any filling bit in the wherein said Decode engine identifying information sequence, and described zero or a plurality of filling bit are not decoded, and the operation of described zero or a plurality of filling bit is used for keeping in fact the amount of the incident of encoding, the relation between coded data piece number and the bit number at information sequence just.
 And

20. arithmetic decoder according to claim 19 is characterized in that:
 described filling bit comprises that the front has one or more byte of paddings of zero or a plurality of alignment bits.

21. arithmetic decoder according to claim 20 is characterized in that:
 described Decode engine is discerned byte of padding by the identification pattern relevant with byte of padding.

22. arithmetic decoder according to claim 19 is characterized in that:
 described Decode engine was decoded to header before the arithmetic decoding coded data.

23. arithmetic decoder according to claim 19 is characterized in that:
 after the last incident of decoded events sequence, do not carry out normalization again.

24. a coding/decoding method comprises:

Produce a plurality of incidents of sequence of events according to the bit stream that comprises coded data;
AndBe identified in any filling bit of following in the bit stream after coded data, and filling bit is not carried out decoding, described filling bit operation is used for keeping in fact the relation of the bit number of the amount of the incident of encoding, the piece number of encoding and bit stream.


25. method according to claim 24 is characterized in that:
 described filling bit comprises that the front has one or more byte of paddings of zero or a plurality of alignment bits.

26. method according to claim 25 is characterized in that:
 the identification byte of padding comprises the pattern that identification is relevant with byte of padding.

27. method according to claim 24 is characterized in that:
 described arithmetic decoder produces sequence of events according to coded data.

28. method according to claim 27 is characterized in that also being included in and uses arithmetic decoder before header to be decoded.

29. method according to claim 27 is characterized in that:
 after the last incident of decoded events sequence, do not carry out normalization again.

30. the manufacture with one or more recordable medias of store instruction on it, described instruction is carried out by system, makes system come decoded data in the following manner:

Produce a plurality of incidents of sequence of events according to the bit stream that comprises coded data;
AndBe identified in any filling bit of following in the bit stream after coded data, and filling bit is not carried out decoding, described filling bit is operated the relation of the bit number that is used for keeping in fact the amount of the incident of encoding, the piece number of encoding and has encoded.


31. method according to claim 30 is characterized in that:
 described filling bit comprises that the front has one or more byte of paddings of zero or a plurality of alignment bits.

32. method according to claim 31 is characterized in that:
 the identification byte of padding comprises the pattern that identification is relevant with byte of padding.

33. method according to claim 30 is characterized in that:
 described arithmetic decoder produces sequence of events according to coded data.

34. method according to claim 30 is characterized in that before also being included in the arithmetic decoding coded data header being decoded.

35. an arithmetic decoder comprises:

Sequencer is used to produce the environmental labels symbol at the incident of sequence of events;
AndThe probability Estimation device is used for determining at the value of LPS with at the probability Estimation of LPS;
AndDecode engine, comprise range registers, be used for numerical value is distributed to scope at LPS, if environmental labels symbol and be not equal to index wherein, then described numerical value is based on the numerical value of being stored in probability Estimation, the range registers and environmental labels symbol, and if the environmental labels symbol equal described index, then this value is not based on the value of being stored in the range registers, and Decode engine also according at the value range of LPS with from the bit of information sequence, is determined the value of binary system incident.


36. arithmetic decoder according to claim 35 is characterized in that described Decode engine stops decoding when environmental labels symbol equals index and LPS decoded.

37. arithmetic decoder according to claim 36 is characterized in that arithmetic coding data not follow after the data of arithmetic coding in information sequence.

38. arithmetic decoder according to claim 36 is characterized in that the ending of described index expression fragment designator.

39. arithmetic decoder according to claim 35, it is characterized in that described Decode engine comprises the numerical value register, and when the environmental labels symbol equals index, if the numerical value in the numerical value register is less than the quantity of distributing to the LPS scope, if value then by producing first state or the value in the numerical value register are more than or equal to described quantity, then, incident is decoded according to the value in the numerical value register by producing the incident of second state.

40.
40., only it is characterized in that the Decode engine response is carried out normalization again to the decoding of incident when the value in the numerical value register during more than or equal to described quantity according to the described arithmetic decoder of claim 39.

41. arithmetic decoder according to claim 35, it is characterized in that described Decode engine comprises the numerical value register, and when the environmental labels symbol equals described index, come incident is decoded by at first from range registers, deducting the numerical value of distributing to LPS, if and the numerical value in the numerical value register is more than or equal to the numerical value in the range registers, generation incident under first state, if perhaps the numerical value in the numerical value register is less than the numerical value in the range registers, then generation incident under second state.

42. according to the described arithmetic decoder of claim 41, only it is characterized in that when the numerical value in the numerical value register during more than or equal to the numerical value in the range registers, the response decoded events, described Decode engine is carried out normalization again.

43.
43., it is characterized in that the numerical value of then distributing to the scope of LPS is 2 if the environmental labels symbol equals described index according to the described arithmetic decoder of claim 41.

44.
44., it is characterized in that the numerical value of then distributing to the scope of LPS is 100Hex if the environmental labels symbol equals described index according to the described arithmetic decoder of claim 42.

45. according to the described arithmetic decoder of claim 41, it is characterized in that if the environmental labels symbol equals described index, the numerical value of then distributing to the scope of LPS is 2, and the last bit that reads in the bit stream of decoded binary system incident equals 1 just comprising.

46. a coding/decoding method comprises:

Produce the environmental labels symbol of the binary system incident of sequence of events; Determine at LPS value and at the probability Estimation of LPS; Numerical value is distributed to scope at LPS, if environmental labels symbol and be not equal to index wherein, then described numerical value accords with the scope of distributing at LPS based on value of being stored in probability Estimation, the range registers and environmental labels, if and environmental labels symbol equals described index, then described numerical value is not based on the numerical value of being stored in the range registers;
AndAccording at the value range of LPS with from the bit of information sequence, determine the value of binary system incident.


47.
47., it is characterized in that also comprising:  when environmental labels symbol equals described index and LPS decoded, stop decoding according to the described method of claim 46.

48.
48., it is characterized in that arithmetic coding data not follow after the data of arithmetic coding in information sequence according to the described method of claim 46.

49.
49., it is characterized in that the ending of described index expression fragment designator according to the described method of claim 46.

50. according to the described method of claim 46, it is characterized in that according to determining that at the value range of LPS with from the bit of information sequence the numerical value of binary system incident comprises:
 when the environmental labels symbol equals described index, if the numerical value in the numerical value register is less than the numerical value of distributing to the LPS scope, if incident then by producing first state or the value in the numerical value register are more than or equal to described quantity, then, incident is decoded according to the value in the numerical value register by producing the incident of second state.

51. according to the described method of claim 50, it is characterized in that also comprising:
 only when the value in the numerical value register during more than or equal to described quantity, response is carried out normalization again to the decoding of incident.

52. according to the described method of claim 46, it is characterized in that according to determining that at the value range of LPS with from the bit of information sequence the value of binary system incident comprises:
 when the environmental labels symbol equals described index, come incident is decoded by at first from range registers, deducting the numerical value of distributing to the LPS scope, if and the numerical value in the numerical value register is more than or equal to the numerical value in the range registers, generation incident under first state, if perhaps the numerical value in the numerical value register is less than the numerical value in the range registers, then generation incident under second state.

53. according to the described method of claim 52, it is characterized in that also comprising:
 only when the numerical value in the numerical value register during more than or equal to the numerical value in the range registers, the response decoded events is carried out normalization again.

54.
54., it is characterized in that the numerical value of then distributing to the scope of LPS is 2 if the environmental labels symbol equals described index according to the described method of claim 52.

55.
55., it is characterized in that the numerical value of then distributing to the scope of LPS is 100Hex if the environmental labels symbol equals described index according to the described method of claim 53.

56. according to the described arithmetic decoder of claim 53, it is characterized in that if the environmental labels symbol equals described index, the numerical value of then distributing to the scope of LPS is 2, and the last bit that reads in the bit stream that comprises the binary system incident of being decoded by Decode engine equals 1.

57. the manufacture with one or more recordable medias of store instruction on it, described instruction is carried out by system, makes system come coded data in the following manner:

Generation is at the environmental labels symbol of binary system incident; Determine at the numerical value of LPS with at the probability Estimation device of LPS; Numerical value is distributed to scope at LPS, if environmental labels symbol and be not equal to index, then described numerical value is based on numerical value of being stored in probability Estimation, the range registers and environmental labels symbol;
And if environmental labels symbol equals described index, then described numerical value is not based on the numerical value of being stored in the range registers;
AndAccording at the value range of LPS with determine the value of binary system incident from the bit of information sequence.


58. an arithmetic encoder comprises:

The probability Estimation device, each incident that is used to produce sequence of events all has the probability Estimation of particular value, and wherein the response of probability Estimation device produces described probability Estimation at the respective environment information of described each incident;
AndThe coding engine that links to each other with the probability Estimation device, be used to respond each incident and corresponding probability Estimation thereof, produce zero or a plurality of bit of information sequence, wherein coding engine is before encoding to the ending of splitting signal, utilization is independent of the constant in subrange interval of the numerical value of range registers, incident is encoded, with the ending of arithmetic coding data in the signalisation information sequence.


59. according to the described encoder of claim 58, it is characterized in that:
 the coding engine that utilizes constant that incident is encoded with the incident in signalisation sequence of events ending can be included in any residue content of low register in the information sequence.

60. according to the described encoder of claim 59, it is characterized in that:
 described coding engine carries out exuberant to any residue content of low register, and the last bit that writes during exuberant is set to equal 1.

61. a method of coding data, described method comprises:

Incident in the coding sequence of events produces coded data;
AndUtilize coded data to produce bit stream, the designator that uses when comprising decoding is encoded with the ending of arithmetic coding data in the expression bit stream.


62. according to the described method of claim 61, it is characterized in that:
 the arithmetic coding data are not followed in bit stream after the arithmetic coding data.

63. according to the described method of claim 61, it is characterized in that:
 the described designator of encoding comprises encodes with the ending of signalisation fragment to incident.

64. according to the described method of claim 63, it is characterized in that:
 the coding incident comprises with the ending of signalisation fragment;
before the ending of splitting signal is encoded, utilize the constant in the subrange interval of the numerical value that is independent of range registers.
 the coding incident comprises with the ending of signalisation fragment;

65.
65., it is characterized in that:  utilize constant to come the coding incident can any residue content of low register is exuberant in information sequence with the ending of signalisation fragment according to the described method of claim 64.

66. according to the described method of claim 65, it is characterized in that:
 any residue content of exuberant low register comprises;
the last bit that writes during exuberant is set to equal 1.
 any residue content of exuberant low register comprises;

67. the manufacture with one or more recordable medias of store instruction on it, described instruction is carried out by system, makes system come coded data in the following manner:

Incident in the coding sequence of events produces coded data;
AndUtilize coded data to produce bit stream, be included in the ending of splitting signal encoded before, utilization is independent of the constant in subrange interval of the numerical value of range registers, and incident is encoded with the ending of arithmetic coding data in the signalisation information sequence.


68.
68., it is characterized in that also comprising instruction according to the described manufacture of claim 67, when carrying out, make the content of the exuberant low register of system by system, comprise that the last bit that is write during exuberant is set to equal 1.

69. the equipment of a coded data, described equipment comprises:

Be used for coded data block to produce the device of coded data;
AndUtilize coded data to produce the device of bit stream, be included in the ending of splitting signal encoded before, utilization is independent of the constant in subrange interval of the numerical value of range registers, and described coding engine is encoded with the device of the ending of arithmetic coding data in the signalisation information sequence to incident.


70. an establishment is used for the method for the state machine of probability Estimation, described method comprises:

Give state in the question blank (LUT) with probability assignments, the maximum probability that the probability that comprises each state worker of these states is set to LPS multiply by i power of fitting percentage, and wherein i is the quantity of given state, and fitting percentage is less than 1; When observing MPS and LPS, produce state transitions at the state that will transfer among the LUT, if wherein current state is not high state, the NextState that state machine is transferred to it from current state when observing MPS is the NextState that is higher than current state, if current state is high state, then described NextState is a current state, in addition, and when the NextState that state machine is transferred to it from current state during to LPS at a plurality of state observation is the version that rounds off of following result of calculation; The probability of current state number+log (the adaptive ratio of the probability * of current state+(the adaptive ratio of 1))/current state)/log (adaptive ratio).


71. according to the described method of claim 70, it is characterized in that:
 described result'"'"'s the version that rounds off is on an average, and any rounding off of being introduced when producing NextState when observing LPS is essentially zero.

72. according to the described method of claim 70, it is characterized in that:
 described LUT has a plurality of values with each state relation, and each in wherein a plurality of values all is similar to the expectation interval range and multiply by product with the probability of state relation.

73. according to the described method of claim 72, it is characterized in that:
 by N/ (2*M*log ((j+M+1)/(j+M))) and with the product of the probability of state relation, and be rounded to integer value and obtain value with state relation, wherein j represents the column index in the array, M is the row number in the array, and N is a constant.

74.
74., it is characterized in that:  , in a plurality of numerical value at least one is cut to predetermined number in a plurality of states at least one according to the described method of claim 72.

75. according to the described method of claim 74, it is characterized in that:
 described predetermined number can have at the most a normalized iterative again during the MPS at coding.

76.
76., it is characterized in that according to the described method of claim 73: 
Status number among the LUT is 63, Adaptive ratio equals 1.0/63 power of 0.5/0.01875, and The maximum probability of LPS is 0.5, and Row in the array number are 4, and The value of the row of first in the array is cut to N/4.


77.
77., it is characterized in that described several N is 512 according to the described method of claim 76.

78. an arithmetic encoder comprises:

The probability Estimation device, each incident that is used to produce sequence of events all has the probability Estimation of particular value, the probability Estimation state machine that the utilization of wherein said probability Estimation device is created in the following manner, the respective environment information of described each incident of response produces probability Estimation, and described mode is; Give the state of lookup table (LUT) with probability assignments, the maximum probability that the probability that comprises each state i of these states is set to LPS multiply by i power of fitting percentage, and wherein i is the quantity of given state, and adaptive ratio is less than 1; When observing MPS and LPS, at the state that will transfer among the LUT, produce state transitions, if wherein current state is not high state, the NextState that state machine is transferred to it from current state when observing MPS is the NextState that is higher than current state, if current state is high state, then described NextState is a current state, in addition, when the NextState that state machine is transferred to it from current state during to LPS at a plurality of state observation be the version that rounds off of following result of calculation; The probability of current state number+log (the adaptive ratio of the probability * of current state+(the adaptive ratio of 1))/current state)/log (adaptive ratio);
AndWith the coding engine that the probability Estimation device links to each other, respond each incident and corresponding probability Estimation thereof, produce zero and a plurality of bit of information sequence.


79. according to the described arithmetic encoder of claim 78, it is characterized in that:
 described result'"'"'s the version that rounds off is on an average, and any rounding off of being introduced when producing NextState when observing LPS is essentially zero.

80. according to the described arithmetic encoder of claim 78, it is characterized in that:
 described LUT has a plurality of values related with each arithmetic encoder, and each in wherein a plurality of values all is similar to the expectation interval range and multiply by product with the probability of state relation.

81. 0 described arithmetic encoder according to Claim 8, it is characterized in that:
 by N/ (2*M*log ((j+M+1)/(j+M))) and with the product of the probability of state relation, and be rounded to integer value and obtain value with state relation, wherein j represents the column index in the array, M is the row number in the array, and N is a constant.

82. 0 described arithmetic encoder according to Claim 8 is characterized in that:
 , in a plurality of numerical value at least one is cut to a number in a plurality of states at least one.

83. 2 described arithmetic encoders according to Claim 8 is characterized in that:
 described quantity can have at the most a normalized iterative again during coding MPS.

84.
84., it is characterized in that according to the described arithmetic encoder of claim 78: 
Status number among the LUT is 63, Adaptive ratio equals 1.0/63 power of 0.5/0.01875, and The maximum probability of LPS is 0.5, and Row in the array number are 4, and The value of the row of first in the array is cut to N/4.


85. 4 described arithmetic encoders according to Claim 8, it is characterized in that:
 described several N are 512.

86. an arithmetic decoder comprises:

The probability Estimation device, each incident that is used to produce sequence of events all has the probability Estimation of particular value, the probability Estimation state machine that the utilization of wherein said probability Estimation device is created in the following manner, response produces probability Estimation at the respective environment information of the incident in the sequence of events, and described mode is; Give the state of lookup table (LUT) with probability assignments, the maximum probability that the probability that comprises each state i of these states is set to LPS multiply by i power of adaptive ratio, and wherein i is the quantity of given state, and adaptive ratio is less than 1; When observing MPS and LPS, at the state that will transfer among the LUT, produce state transitions, if wherein current state is not high state, the NextState that state machine is transferred to it from current state when observing MPS is the NextState that is higher than current state, if current state is high state, then described NextState is a current state, in addition, when the NextState that state machine is transferred to it from current state during to LPS at a plurality of state observation be the version that rounds off of following result of calculation; The probability of current state number+log (the adaptive ratio of the probability * of current state+(the adaptive ratio of 1))/current state)/log (adaptive ratio);
AndWith the Decode engine that the probability Estimation device links to each other, respond its corresponding probability Estimation and information sequence, produce the incident of sequence of events.


87. 6 described arithmetic decoder according to Claim 8, it is characterized in that:
 described result'"'"'s the version that rounds off is on an average, and any rounding off of being introduced when producing NextState when observing LPS is essentially zero.

88. 6 described arithmetic decoder according to Claim 8, it is characterized in that:
 described LUT has a plurality of values related with each arithmetic encoder, and each in wherein a plurality of values all is similar to the expectation interval range and multiply by product with the probability of state relation.

89. 8 described arithmetic decoder according to Claim 8, it is characterized in that:
 by N/ (2*M*log ((j+M+1)/(j+M))) and with the product of the probability of state relation, and be rounded to integer value and obtain value with state relation, wherein j represents the column index in the array, M is the row number in the array, and N is a constant.

90. 8 described arithmetic decoder according to Claim 8 is characterized in that:
 , in a plurality of numerical value at least one is cut to a number in a plurality of states at least one.

91. according to the described arithmetic decoder of claim 90, it is characterized in that:
 described quantity can have at the most a normalized iterative again during the MPS at coding.

92. 6 described arithmetic decoder according to Claim 8 is characterized in that:

Status number among the LUT is 63, Adaptive ratio equals 1.0/63 power of 0.5/0.01875, and The maximum probability of LPS is 0.5, and Row in the array number are 4, and The value of the row of first in the array is cut to N/4.


93. according to the described arithmetic encoder of claim 92, it is characterized in that:
 described several N are 512.

94. a coding/decoding method comprises:

Each incident that produces sequence of events all has the probability Estimation of particular value, utilizes the probability Estimation state machine of creating in the following manner, and response produces described probability Estimation at the respective environment information of the incident in the sequence of events, and described mode is; Give the state of lookup table (LUT) with probability assignments, the maximum probability that the probability that comprises each state i of these states is set to LPS multiply by i power of adaptive ratio, and wherein i is the quantity of given state, and adaptive ratio is less than 1; When observing MPS and LPS, produce state transitions at the state that will transfer among the LUT, if wherein current state is not high state, then the NextState that state machine is transferred to it from current state when observing MPS is the NextState that is higher than current state, if current state is high state, then described NextState is a current state, in addition, and when the NextState that state machine is transferred to it from current state during to LPS at a plurality of state observation is the version that rounds off of following result of calculation; The probability of current state number+log (the adaptive ratio of the probability * of current state+(the adaptive ratio of 1))/current state)/log (adaptive ratio);
AndRespond its corresponding probability Estimation and information sequence, produce the incident of sequence of events.

Specification(s)