WO2011113290A1

WO2011113290A1 - Method for processing arithmetic entropy coding in real time at high-speed based on joint photographic experts group (jpeg) 2000 standard

Info

Publication number: WO2011113290A1
Application number: PCT/CN2010/080584
Authority: WO
Inventors: 刘凯; 王柯俨; 李云松; 曾伟
Original assignee: 西安电子科技大学
Priority date: 2010-03-19
Filing date: 2010-12-31
Publication date: 2011-09-22

Abstract

A method for processing arithmetic entropy coding in real time at a high-speed based on the Joint Photographic Experts Group (JPEG) 2000 standard is disclosed. The method comprises: a probability interval value determining step, judging a coding symbol type according to two context symbol pairs inputted in parallel, and updating probability interval values and coding parameters according to the coding symbol type; a code value normalization step, updating code values according to the coding symbol type, judging current normalization type and performing normalization according to the updated probability interval values and coding parameters, and outputting a normalization code stream and an emptying code stream; a code stream buffering step, in accordance with the order of code stream output, classifying and buffering the normalization code stream and the emptying code stream respectively according to an amount of output bytes of the normalization code stream and the emptying code stream, and outputting the normalization code stream and the emptying code stream in parallel; and an output step, according to a prescribed priority order, outputting the normalization code stream and the emptying code stream outputted in parallel at the code stream buffering step in serial and in turn.

Description

High-speed real-time processing arithmetic entropy coding method based on Joint Photographic Experts Group (JPEG) 2000 standard

Technical field

The invention relates to a high-speed real-time processing arithmetic entropy coding method based on the JPEG2000 standard, which is used for image compression coding of various digital devices, in particular high-speed real-time satellite image compression coding. Background technique

With the development of multimedia and network technologies and their applications in medical imaging, remote sensing imagery and digital image/video transmission, the existing still image compression standard JPEG can no longer meet the requirements of current practical applications, for which the International Organization for Standardization In November 2000, a new still image compression standard JPEG2000 was developed. One of the core technologies of the standard is to use the arithmetic entropy coding method to encode the wavelet transformed data to achieve image data compression processing.

The basic encoding process of JPEG2000 can be simply described as follows: First, the input image is subjected to discrete wavelet transform, and the wavelet transformed data is called wavelet coefficient; then the wavelet coefficients are quantized and encoded in units of code blocks, and the size of the code block is usually 32x32. Or 64x64; In the encoding process, the corresponding context model is established according to the wavelet coefficient bit plane order, and the context symbol pair CXD is encoded by the arithmetic entropy coding method, thereby obtaining the coded data of the corresponding code block; finally, according to the preset compression ratio The coded data corresponding to each code block is selectively outputted, and the encoding process of the single image is completed.

The coding principle of the arithmetic entropy coding method is described as follows:

The input of the arithmetic entropy coding is a context symbol pair CXD, including a context label CX and a context decision D, where CX represents a context label generated by the current coded pixel, and the value ranges from 0 to 18, and D represents a decision of the corresponding context, and the value range is It is 0-1. The output of the arithmetic entropy encoding of the context symbol pair is the corresponding compressed code stream. In the arithmetic entropy coding, the probability of the corresponding decision D is adaptively selected according to the input context label CX, and the symbol values to be encoded according to the judgment D are prepared, that is, the high probability coding symbol MPS and the small probability coding symbol LPS, and small Probability encodes the probability Qe corresponding to the symbol LPS, then adjusts the corresponding probability interval value, and outputs the encoded value to the code value register. The probability interval value is represented by a 16-bit unsigned integer, which is 32 bits and is divided into 5 different fields, that is, bits 31 to 28 represent 4-bit zero data, and bit 27 represents "into Bit "bit, bits 26 to 19 indicate the coded output bits, bits 18 to 16 indicate the 3-bit division bits used to block the carry propagation, and bits 15 to 0 represent the decimal places of the code value. For the probability Qe of the coded symbols, The arithmetic entropy encoder is implemented by an array of 47 indexes, which is called a probability estimation table PET, and each item in the table corresponds to a Qe value of a 16-bit LPS symbol. According to the JPEG2000 standard, the current coding symbol is corresponding. The context label CX serves as an address to access an array of size 19 X 6 to obtain an entry index Index of the probability estimation table PET, and then reads the corresponding LPS symbol probability value Qe in the PET table by using the Index as the address, and combines the current probability interval value. And the code value determines the specific coding process. If the current probability interval value is less than or equal to 0.75, the value corresponds to 0x8000 in hexadecimal. In order to prevent overflow, the probability interval value and the code value need to be normalized, by left The probability interval value and the code value make the probability interval larger than 0x8000. At the same time of normalization, if the code stream forms a whole byte in the code value, the code stream output process is performed, that is, Final encoded output stream. To prevent carry propagation, by setting specific bits are filled into the transfer position to intercept.

As the arithmetic coding method as described above, for example, the following technique is disclosed:

(1) Chinese Taiwan scholars K.-K. Ong and Jen-Shiun Chiang, respectively, at the 2002 Int. Conf. Image Process. He Wei 2004 IEEE Int. Symp. Circuits and Systems International Conference "A high throughput" Low cost context-based adaptive arithmetic codec for multiple standards "(2002, vol.1, pp.1872 - 1875) and "High-speeds EBCOT with dual context- modeling coding architecture for JPEG2000" (2004, vol. 3, pp. The high-speed arithmetic entropy coding method proposed in 865-868.), wherein the update process of the probability interval value and the normalization of the code value are performed according to a single context symbol pair, so that the throughput rate is one context symbol for a single clock. Correct.

(2) The article "VLSI Design of a High-Speed and Area-Efficient JPEG2000 Encoder" published by domestic scholar Mei Kuizhi and others in the IEEE Trans, on Circuits and System for Video Technology 2007 (Vol. 17, 2007, The arithmetic entropy coding method given in the eighth issue, pages 1065-1078, in which the arithmetic entropy coding is implemented by the multi-input synchronous pipeline technique, but the update of the probability interval value and the code value are also based on a single context symbol pair. Normalized processing, the overall processing speed is only 0.625 context symbol pairs processed by a single clock, and the encoding speed is low.

(3) Domestic expert Xu Chao's article at the 2005 Conf. Image Process. (ICIP'05) International Conference "A Dual-Symbol Coding Arithmetic Coder Architecture Design For High Speed EBCOT Coding Engine in JPEG2000 "The arithmetic entropy coding method is given. When the context labels of two context symbol pairs are different, the update processing of the probability interval value and the normalization processing of the code value reach a single clock processing. Two context symbol pairs, but when the context labels of the two context symbol pairs are the same, the update of the probability interval values of the two context symbol pairs cannot be processed in parallel, so that the subsequent normalization of the code values can only be processed serially. This caused a pause in the entire encoding process, and its processing power is only one context symbol pair per clock.

(4) Australian scholar M. Dyer published in 2006 IEEE Transactions on Circuits and Systems-I: Regular Papers article "Concurrency Techniques for Arithmetic Coding in JPEG2000" (vol. 53, no. 6, pp. 1203- The high-speed arithmetic entropy coding method described in 1213, June 2006), which can implement parallel processing of dual context symbol pairs, but requires more internal memory bits, which poses certain difficulties for chip design.

As described above, the conventional arithmetic entropy coding method has the following problems: If the implementation complexity is low, the processing speed is also low; if the processing speed is increased, the implementation complexity is drastically increased. Summary of the invention

The present invention has been made in view of the above problems, and an object thereof is to avoid the above-mentioned deficiencies of the prior art, and to provide a high-speed real-time processing arithmetic entropy coding method based on the JPEG2000 standard, so as to improve the coding system while maintaining the same complexity. Processing speed.

In order to achieve the above object, a high-speed real-time processing arithmetic entropy coding method based on the JPEG2000 standard of the present invention includes: a probability interval value determining step of determining a coded symbol type according to two context symbol pairs input in parallel, and according to the Encoding symbol type, updating the probability interval value and the encoding parameter; the code value normalization step, updating the code value according to the above coding symbol type, determining the current normalization type according to the updated probability interval value and the coding parameter Performing normalization processing, and outputting the normalized code stream and the empty code stream; the code stream buffering step, according to the order of output of the code stream, according to the outputted byte number of the normalized code stream and the empty code stream And respectively classifying and buffering the normalized code stream and the empty code stream, and outputting in parallel; and outputting, according to the normalized code stream and the empty code stream output in parallel in the code stream buffer step of

DRAWINGS 1 is a flow chart of an arithmetic entropy coding method of the present invention.

Fig. 2 is a detailed flowchart of the probability interval value determining process of the present invention.

3 is a flow chart of the double-probability symbol probability interval value update subroutine of the present invention.

4 is a flow chart of the high probability/small probability symbol probability interval value update subroutine of the present invention.

Figure 5 is a flow chart showing the small probability/high probability symbol probability interval value update subroutine of the present invention.

6 is a flow chart of the double small probability symbol probability interval value update subroutine of the present invention.

Figure 7 is a detailed flow chart of the normalization method of the present invention.

Figure 8 is a flow chart of the double-high probability symbol code value normalization type judgment sub-routine of the present invention.

Fig. 9 is a flow chart showing the judgment of the large probability/small probability symbol code value normalization type of the present invention.

Fig. 10 is a flow chart showing the small probability/high probability symbol code value normalization type judgment sub-render of the present invention.

Figure 11 is a flow chart of the single normalization process of the present invention.

Figure 12 is a flow chart of a type of double normalization process of the present invention.

Figure 13 is a flow chart of the second-class double normalization process of the present invention.

14 is a detailed flowchart of the operation of the classification buffer output code stream of the present invention;

Figure 15 is a detailed flow chart showing the operation of the code stream of the output classification buffer of the present invention. detailed description

The encoding method of the present invention is implemented on Xilinx's model XC2V3000-6BG728 using Xilinx ISE 9.1 integrated development software and VHDL, Verilog HDL language.

The high-speed real-time processing arithmetic entropy coding method based on the JPEG2000 standard of the present invention will be described below with reference to FIG. 1 is a flow chart of the above arithmetic entropy encoding method.

As shown in FIG. 1, the arithmetic coding method of the present invention includes: Step S1, determining a type of a double context symbol pair; Step S2, updating a probability interval value; Step S3, normalizing the code value; Step S4: The code value is subjected to the emptying process; step S5, the classifying buffer outputs the code stream; and step S6: and outputting the code stream of the classifying buffer. Step S1 and step S2 constitute a probability interval value determining step, and step S3 and step S4 form a code value normalization step.

Hereinafter, the operation of each of the above steps will be described with reference to Figs. 2 to 15 .

(probability interval value determination)

Step S1, determining the type of the double context symbol pair.

Step S11: reading the double-encoded index and the double-large probability symbol identifier. Simultaneously inputting a first context symbol pair (CX0, DO) and a second context symbol pair (CX1, D1), respectively, using the first context label CX0 and the second context label CX1 as addresses, from a preset array of encoded indices And reading a corresponding first coding index Index0 and a second coding index Index1, and a first large probability symbol identifier MPS_CX0 and a second large probability symbol identifier MPS_CX1 respectively corresponding to the first context label CX0 and the second context label CX1. The encoded index array has a capacity of 19x7 bits. CX0 and CX1 have a value range of 0-18, represented by a 5-bit binary bit string. IndexO and Indexl range from 0 to 46, represented by a 6-bit binary string. MPS_CX0 and MPS_CX1 have a value range of 0-1. It is represented by a 1-bit binary bit string.

Step S12: Determine the probability type of the pair of double context symbols.

The probability of the two context symbol pairs is determined by the relationship between the two large probability symbols obtained in step S11 and the input two context decisions D0, D1. The probability types include the double probability symbol MPSMPS, the large probability/small probability symbol MPSLPS, the small probability/high probability symbol LPSMPS and the double small probability symbol LPSLPS, which are represented by a 2-bit binary bit string.

If the first large probability symbol identifier MPS_CX0 is equal to the first decision DO, and the second approximate rate symbol identifier MPS_CX1 is equal to the second decision D1, the probability type of the double context symbol pair is determined as a double large probability symbol MPSMPS;

If the first large probability symbol identifier MPS_CX0 is equal to the first decision D0, and the second large probability symbol identifier MPS_CX1 is not equal to the second decision D1, the probability type of the double context symbol pair is determined to be a large probability/small probability symbol MPSLPS;

If the first large probability symbol identifier MPS_CX0 is not equal to the first decision D0, and the second large probability symbol identifier MPS_CX1 is equal to the second decision D1, the probability type of the double context symbol pair is determined to be a small probability/high probability symbol LPSMPS ;

If the first large probability symbol identification MPS_CX0 is not equal to the first decision D0, and the second large probability symbol identification MPS_CX1 is not equal to the second decision D1, the probability type of the double context symbol pair is determined to be the double small probability symbol LPSLPS.

Step S13: Determine the similarity and difference of the context labels according to the input double context label.

If the first context label CX0 is equal to the second context label CX1, it is determined that the double context labels are the same, otherwise, it is determined that the double context labels are different. The similarities and differences of the double context labels are identified by the distinct label DIFF. The range of DIFF is 0-1, which is represented by a 1-bit binary string. If the double context labels are the same, set the distinct flag DIFF to 0, otherwise set to 1. Step S14: According to the similarity and the same type of the context label and the probability type of the double context symbol pair, the type of the double context symbol pair is determined as follows.

(1) If the two context labels are the same and the probability type of the double context symbol pair is a double probability symbol, the double context symbol pair type is the same as the double large probability symbol;

(2) If the two context labels are the same, and the probability type of the double context symbol pair is a large probability/small probability symbol, the double context symbol pair type is the same as the large probability/small probability symbol;

(3) If the two context labels are the same and the probability type of the double context symbol pair is a small probability/large probability symbol, the double context symbol pair type is the same as the small probability/large probability symbol;

(4) If the two context labels are the same, and the probability type of the double context symbol pair is a double small probability symbol, the double context symbol pair type is the same as the double small probability symbol;

(5) If the two context labels are different, and the probability type of the double context symbol pair is a double-probability symbol, the double-context symbol pair type is double-probability symbol different;

(6) If the two context labels are different, and the probability type of the double context symbol pair is a large probability/small probability symbol, the double context symbol pair type is a large probability/small probability symbol different;

(7) If the two context labels are different, and the probability type of the double context symbol pair is a small probability/large probability symbol, the double context symbol pair type is a small probability/large probability symbol different;

(8) If the two context labels are different and the probability type of the double context symbol pair is a double small probability symbol, the double context symbol pair type is a double small probability symbol.

Step S2: The probability interval value is updated (step S15 to step S17).

Obtaining two corresponding small probability symbol probability values from the set probability estimation table according to the double coding index obtained in step S11 (step S15), and obtaining the step probability information according to the obtained small probability symbol probability value. The eight types of double context symbol pairs are respectively subjected to probability interval value updating, and the encoding parameters are calculated, the encoding parameters including two probability interval value shift count values and a normalization process identifier (step S16). Finally, the updated probability interval value is output (step S17). Thereby, the determination of the probability interval value is ended.

Next, the probability interval value update processing will be described in detail.

(A) designing a specific probability estimate table PET, according to the first coding index IndexO and the second coding index Index1 obtained in step S1, respectively obtaining the corresponding first small probability symbol probability value QeO and the first from the probability estimation table Two small probability symbol probability value Qel.

The probability evaluation table PET designed by the present invention is composed of four sub-tables, which are: (a) a small probability symbol probability value table for storing a small probability symbol probability value Qe;

(b) a probability value table in which the next context symbol is a large probability symbol, and is used to store a probability value of the next upper and lower symbol as a large probability symbol NT_NMPS_QE1;

(c) a probability value table in which the next context symbol is a small probability symbol, and is used to store a probability value of the next upper and lower text symbol as a small probability symbol NT_NLPS_QE1;

(d) a preamble zero count value table for storing the leading zero count value of the small probability symbol probability value, that is, 16 minus all possible differences of the most significant bit position of the small probability symbol probability value, ranging from 1 to 15.

According to the JPEG2000 standard, the probability in each probability table is represented by a 16-bit binary number, but since the lowest two bits of the probability value given in the standard are "01", and the highest bit is "0". ", therefore, 13 bits can be used for storage, instead of 16 bits to represent a single probability value. In addition, although the JPEG2000 standard specifies that each probability value table stores probability values corresponding to 47 indexes, there are cases where the probability values are the same, such as values of Qe [0], Qe [6], Qe [14], Qe [46]. Both are hexadecimal values 0x5601, so only different probability values need to be stored during storage, and each memory cell is located by a simple operation.

For the probability table of the small probability symbol probability value table and the next context symbol for the large probability symbol, since there are only 32 different probability values in the two tables, the depth of both tables is 32, and the width is 13 bits, capacity is 32x13 bits. For the probability table whose next context symbol is a small probability symbol, since there are only 30 different probability values in the table, the table has a depth of 30, a width of 13 bits, and a capacity of 30x13 bits. For the leading zero count table, since each leading zero count value needs to be represented by 4 bits, the table has a depth of 32, a width of 4 bits, and a capacity of 32x4 bits.

Taken together, the total capacity of the probability estimate table is 32x13 + 32x13 + 30x13 + 32x4 = 1350 bits. The specific contents of the probability valuation table are shown in Table 1, Table 2, and Table 3, respectively.

Small probability symbol probability value subtable and leading zero count subtable

6

l78S080/0l0ZN3/X3d Οτ

褂索褂, "/^ 3⁄4:^丄丁-丄ε3⁄4

l78S080/0l0ZN3/X3d 21 0xA8

22 0x88

23 0x50

24 0x44

25 0x21

26 0x12

27 0x9

28 0x5

29 0x2

(B) According to the type of double context symbol pair, the probability interval value update is divided into four different cases, including double large probability symbol probability interval value update, large probability/small probability symbol probability interval value update, small probability/high probability Symbol probability interval value update and double small probability symbol probability interval value update. For each case, the current probability interval values are updated separately, and the detailed process is shown in Figure 3 to Figure 6:

(a) The process of updating the probability interval value of the double probability symbol is as follows:

As shown in FIG. 3, first, the first, second, and third temporary variables are established, and the first small probability symbol probability value QeO and the second small probability symbol probability value Qel are respectively saved to the first temporary variable temp2 and the second In the temporary variable tempO, the difference between the current probability interval value A and the first small probability symbol probability value QeO, that is, A-Qe0, is simultaneously calculated and saved in the third temporary variable tempi;

Secondly, the first, second and third temporary variable values are respectively updated, wherein: for the first temporary variable temp2, if the first temporary variable value is smaller than the hexadecimal value 0x8000, the first temporary variable value is shifted to the left One bit, the process is repeated thereafter until the first temporary variable value is greater than or equal to the hexadecimal value 0x8000; for the second temporary variable tempO, if the distinct tag value is 0, the readout is performed from the probability estimate table A context symbol is a probability value NT_NMPS_QE1 of a large probability symbol, and the second temporary variable value is updated to NT_NMPS_QE1 _; for the third temporary variable tempi, the update process is the same as the first temporary variable temp2;

Again, the probability interval temporary variable value A_temp is calculated according to the size of the probability interval value A: If the probability interval value A is greater than or equal to 2xQeO, the relationship between the third temporary variable tempi and the second temporary variable tempO is compared, if tempi is less than 2xtemp0, Bay J A_temp Equivalent to tempO, otherwise, A_temp is equal to templ-tempO; If the probability interval value A is less than 2><QeO, the relationship between the first temporary variable temp2 and the second temporary variable tempO is compared, and if temp2 is less than 2xtemp0, then A_temp Equal to tempO, otherwise, A—temp is equal to temp2-temp0;

Then, the probability interval temporary variable value A_temp is corrected. If the probability interval temporary variable value A_temp is smaller than the hexadecimal value 0x8000, the probability interval temporary variable value is shifted to the left by one bit, and the judgment process is repeated until the probability The interval temporary variable value is greater than or equal to the hexadecimal value 0x8000; Finally, the probability interval value A is updated, and the probability interval value A is updated to the corrected probability interval temporary variable value A_temp.

(b) Update the probability/interval probability interval values for large probability/small probability symbols as follows:

As shown in FIG. 4, first, the first, second, and third temporary variables are established, and the first small probability symbol probability value QeO and the second small probability symbol probability value Qel are respectively saved to the first temporary variable temp2 and the second In the temporary variable tempO, the difference between the current probability interval value A and the first small probability symbol probability value QeO, that is, A-Qe0, is simultaneously calculated and saved in the third temporary variable tempi;

Secondly, the first, second and third temporary variable values are respectively updated, wherein: for the first temporary variable temp2, if the first temporary variable value is smaller than the hexadecimal value 0x8000, the first temporary variable value is shifted to the left One bit, the process is repeated thereafter until the first temporary variable value is greater than or equal to the hexadecimal value 0x8000; for the second temporary variable tempO, if the distinct tag value is 0, the readout is performed from the probability estimate table A context symbol is a probability value NT_NMPS_QE1 of a large probability symbol, and the second temporary variable value is updated to NT_NMPS_QE1; for the third temporary variable tempi, the update process is the same as the first temporary variable temp2;

Again, the probability interval temporary variable value A_temp is calculated according to the size of the probability interval value A _: if the probability interval value A is greater than or equal to 2xQeO, the relationship between the third temporary variable tempi and the second temporary variable tempO is compared, and if tempi is greater than or equal to 2xtemp0, then A—temp is equal to tempO, otherwise, A—temp is equal to templ-tempO; if the probability interval value A is less than 2xQeO, the relationship between the first temporary variable temp2 and the second temporary variable tempO is compared. If temp2 is greater than or equal to 2xtemp0, then A_temp is equal to tempO Otherwise, A_temp is equal to temp2-temp0;

Then, the probability interval temporary variable value A_temp is corrected. If the probability interval temporary variable value A_temp is smaller than the hexadecimal value 0x8000, the probability interval temporary variable value is shifted to the left by one bit, and the judgment process is repeated until the probability The interval temporary variable value is greater than or equal to the hexadecimal value 0x8000;

Finally, the probability interval value A is updated, and the probability interval value A is updated to the corrected probability interval temporary variable value A_temp.

(c) The process of updating the probability interval value of the small probability/high probability symbol is as follows: As shown in FIG. 5, first, the first, second, and third temporary variables are established, and the first small probability symbol probability value QeO and the second small probability symbol probability value Qel are respectively saved to the first temporary variable temp2 and the second In the temporary variable tempO, the difference between the current probability interval value A and the first small probability symbol probability value QeO, that is, A-Qe0, is simultaneously calculated and saved in the third temporary variable tempi;

Secondly, the first, second and third temporary variable values are respectively updated, wherein: for the first temporary variable temp2, if the first temporary variable value is smaller than the hexadecimal value 0x8000, the first temporary variable value is shifted to the left One bit, the process is repeated thereafter until the first temporary variable value is greater than or equal to the hexadecimal value 0x8000; for the second temporary variable tempO, if the distinct tag value is 0, the readout is performed from the probability estimate table A context symbol is a probability value NT_NLPS_QE1 of a small probability symbol, and the second temporary variable value is updated to NT_NLPS_QE1; for the third temporary variable tempi, the update process is the same as the first temporary variable temp2;

Again, the probability interval temporary variable value A_temp is calculated according to the magnitude of the probability interval value A. If the probability interval value A is less than 2xQeO, the relationship between the third temporary variable tempi and the second temporary variable tempO is compared, if tempi is less than 2xtemp0, the shell I" A_temp is equal to tempO, otherwise, A_temp is equal to templ-tempO; if the probability interval value A is greater than or equal to 2><QeO, the relationship between the first temporary variable temp2 and the second temporary variable tempO is compared, if temp2 is less than 2 <temp0, Bay lj A – temp is equal to tempO, otherwise A_temp is equal to temp2-temp0;

(d) The double small probability symbol probability interval value update process is as follows:

As shown in FIG. 6, first, the first, second, and third temporary variables are established, and the first small probability symbol probability value QeO and the second small probability symbol probability value Qel are respectively saved to the first temporary variable temp2 and the second In the temporary variable tempO, the difference between the current probability interval value A and the first small probability symbol probability value QeO, that is, A-Qe0, is simultaneously calculated and saved in the third temporary variable tempi;

Secondly, the first, second and third temporary variable values are respectively updated, wherein: for the first temporary variable temp2, if the first temporary variable value is smaller than the hexadecimal value 0x8000, the first temporary variable value is shifted to the left One bit, after which the process is repeated until the value of the first temporary variable is greater than or equal to sixteen For the second temporary variable tempO, if the distinct tag value is 0, the next context symbol is read from the probability estimate table as the probability value NT_NLPS_QE1 of the small probability symbol, and the second is The temporary variable value is updated to NT_NLPS_QE1; for the third temporary variable tempi, the update process is the same as the first temporary variable temp2;

Again, the probability interval temporary variable value A_temp is calculated according to the size of the probability interval value A. If the probability interval value A is less than 2xQeO, the relationship between the third temporary variable tempi and the second temporary variable tempO is compared, if tempi is greater than or equal to 2 X tempO , then A_temp is equal to tempO, no shell lj, A_temp is equal to templ-tempO; if the probability interval value A is greater than or equal to 2 <QeO, the relationship between the first temporary variable temp2 and the second temporary variable tempO is compared, if temp2 is greater than or equal to 2><temp0, then A-temp is equal to tempO, no shell lj, A-temp is equal to temp2-temp0;

(C) Calculating two probability interval value shift count values, wherein the first probability interval value shift count value is a leading zero count value before the probability interval value A is updated, and the second probability interval value shift count value is a probability interval value A updated leading zero count value.

Finally, the updated probability interval value is output.

As described above, the probability estimation table of the present invention fully considers the relationship between the fields in the probability estimation table, and its capacity is only 1350 bits, whereas the probability estimation table adopted by M. Dyer in the prior art requires 47 X 32. X 2 = 3008 bits. In the prior art, the probability estimate table adopted by Liu Qiwei requires 47 x 96 = 4512 bits, which shows that the present invention significantly reduces the internal storage bits.

(Normalized)

Hereinafter, the normalization steps of steps S3 and S4 of Fig. 1 will be described in detail with reference to Fig. 7. Step S3, normalizing the code values.

In this process, according to the double context symbol pair type obtained in step S1.4 of FIG. 2, the code values are respectively updated correspondingly; according to the probability interval value, the normalization process identifier and the small probability symbol probability value, the current return is determined. a normalization type, including zero normalization, single normalization, one type of double normalization, and two types of double normalization; for zero-order normalized type code values, Normalization Rational, there is no corresponding code stream output; for single normalization, one type of double normalization and two types of double normalization type code values, respectively, normalization processing, and generate corresponding single normalization The code stream, a type of double normalized code stream and a second type of double normalized code stream. details as follows:

Step S31: Read the necessary parameters of the encoding

The coding necessary parameters include: a probability interval value A, a normalization process identifier RenormTag, a first probability interval value shift count value NumSLA0, a second probability interval value shift count value NumSLAl, a first small probability symbol probability value QeO, and a second Small probability symbol probability value Qel. Wherein, the normalization process identifier RenormTag represents the probability type of the pair of double context symbols, that is, 0 indicates a double small probability symbol, 1 indicates a double large probability symbol, 2 indicates a large probability/small probability symbol, and 3 indicates a small probability/large probability symbol.

Step S32: Update the code value, and determine the code value left shift count value CT

RenormTag is identified according to the input normalization process. If the first context symbol pair is a large probability symbol, that is, the RenormTag is 1 or 2, the sum of the second small probability symbol probability value Qel and the current code value C is calculated, and the code is The value C is updated to the sum value; otherwise, the code value C is kept unchanged;

And, determining the code value left shift count value CT in the following manner, that is, if the code value normalization process is performed on the context symbol pair for the first time, the CT is assigned according to the initialization method given by the JPEG2000 standard, otherwise, the CT is maintained. The value does not change.

Step S33: determining the current normalization type

Step S331: If the normalization process identifier RenormTag is 1, the double-probability symbol code value normalization type judgment is performed. The detailed process is shown in Figure 8.

Step S3311: If the probability interval value A is less than twice the first small probability symbol probability value QeO, step S3312 is performed; otherwise, step S3314 is performed;

Step S3312: Adjusting the first small probability symbol probability value QeO, that is, shifting QeO to the left until QeO is greater than or equal to the hexadecimal value 0x8000, if the adjusted first small probability symbol probability value SHIFT_V(Qe0) is smaller than The second small probability symbol probability value Qel is twice, then the normalization type is a type of double normalization, and a type of double normalization type indication is set, otherwise, step S3313 is performed; step S3313: if the adjustment The difference between the first small probability symbol probability value SHIFT_V(Qe0) and the second small probability symbol probability value Qel is less than the hexadecimal value 0x8000, then the normalization type is the second class double normalization, and Set two types of double normalization type indication; otherwise, the normalization type is single normalization, and set a single normalization type indication; Step S3314: If the difference between the probability interval value A and the first small probability symbol probability value QeO is less than the hexadecimal value 0x8000, step S3315 is performed; otherwise, step S3317 is performed;

Step S3315: Adjusting the difference between the probability interval value A and the first small probability symbol probability value QeO, if the adjusted difference SHIFT_V (A-QeO) is less than twice the first small probability symbol probability value QeO, then returning The normalization type is a type of double normalization, and a type of double normalization type indication is set; otherwise, step S3316 is performed;

Step S3316: If the adjusted difference SHIFT_V (A-QeO) is smaller than the sum of the second small probability symbol probability value Qel and the hexadecimal value 0x8000, the normalization type is the second class double normalization, and is set. The second type of double normalization type indication; otherwise, the normalization type is single normalization, and a single normalization type indication is set;

Step S3317: If the difference between the probability interval value A and the first small probability symbol probability value QeO is less than twice the second small probability symbol probability value Qel, the normalization type is a single normalization, and a single return is set. a type indication; otherwise, step S3318;

Step S3318: If the difference between the probability interval value A and the first small probability symbol probability value QeO is smaller than the sum of the second small probability symbol probability value Qel and the hexadecimal value 0x8000, the normalization type is a single normalization. , and set a single normalization type indication; otherwise, the normalization type is zero normalization, and set zero normalization type indication.

Step S332: If the normalization process identifier RenormTag is 2, the large probability/small probability symbol code value normalization type judgment is performed. The detailed process is shown in Figure 9.

Step S3321: If the probability interval value A is less than twice the first small probability symbol probability value QeO, then step S3322 is performed; otherwise, step S3323 is performed;

Step S3322: Adjust the first small probability symbol probability value QeO, that is, shift QeO to the left until QeO is greater than or equal to the hexadecimal value 0x8000. If the adjusted first small probability symbol probability value SHIFT_V(Qe0) is less than twice the second small probability symbol probability value Qel, the normalization type is a second-class double-order normalization, and the second-class double-order is set. Normalized type indication; otherwise, the normalized type is a type of double normalization, and a type of double normalization type indication is set;

Step S3323: If the difference between the probability interval value A and the first small probability symbol probability value QeO is less than the hexadecimal value 0x8000, step S3324 is performed; otherwise, the normalization type is single normalization, and a single return is set. a type indication;

Step S3324: Performing a difference between the probability interval value A and the first small probability symbol probability value QeO Adjustment, if the adjusted difference SHIFT_V(A-QeO) is less than twice the first small probability symbol probability value QeO, the normalization type is a class of double-order normalization, and a class of double-return is set. The type indication; otherwise, the normalization type is the second-class double normalization, and the second-class double-normalization type indication is set.

Step S333: If the normalization process identifier RenormTag is 3, the small probability/high probability symbol code value normalization type judgment is performed. The detailed process is shown in Fig. 10.

Step S3331: If the probability interval value A is less than twice the first small probability symbol probability value QeO, step S3332 is performed; otherwise, step S3334 is performed;

Step S3332: If the difference between the probability interval value A and the first small probability symbol probability value QeO is adjusted to be less than twice the second small probability symbol probability value Qel, the normalization type is a class of double normalization, And setting a type of double normalization type indication; otherwise, performing step S3333;

Step S3333: Adjust the difference between the probability interval value A and the first small probability symbol probability value QeO, if the adjusted difference SHIFT_V (A-QeO) is smaller than the second small probability symbol probability value Qel and hexadecimal value For the sum of 0x8000, the normalization type is the second-class double-normalization, and the second-class double-normalization type indication is set; otherwise, the normalization type is single normalization, and a single normalization is set. Type indication

Step S3334: Adjusting the first small probability symbol probability value QeO, if the adjusted first small probability symbol probability value SHIFT_V(Qe0) is less than twice the second small probability symbol probability value Qel, the normalization type For the second-class double-order normalization, and set the second-class double-substitution type indication; otherwise, perform step S3335;

Step S3335: If the difference between the adjusted first small probability symbol probability value SHIFT_V(Qe0) and the second small probability symbol probability value Qel is less than the hexadecimal value 0x8000, the normalization type is a class of double times Normalize, and set two types of double normalization type indication; otherwise, the normalization type is a type of double normalization, and set a single normalization type indication.

The single normalization type indication in the above step indicates that the code value of the dual context symbol pair is only normalized once, and a type of double normalization type indication and a second type of double normalization type indication indicate that The code value of the double context symbol pair will be normalized twice, and the zero normalization type indication indicates that the code value of the double context symbol pair is not normalized. These four normalized type indications are used to select the corresponding normalization process.

Step 34: Normalize according to the normalized type.

According to the normalization type obtained in step 33, the single normalization type and the first-order double normalization are respectively The code values under the type and the second-order double normalization type are subjected to corresponding normalization processing, and a corresponding normalized code stream is generated.

If the normalization type is single normalization, the code value is subjected to a single normalization process, and the reference figure

5, the specific implementation is as follows:

Step 341: If the first probability interval value shift count value NumSLAO or the second probability interval value shift count value NumSLAl is smaller than the code value left shift count CT, the code value C is shifted to the left by NumSLAO or NumSLAl times, and there is no corresponding single-time normalization. Streaming the stream output, ending the current normalization process; otherwise, performing step 342;

Step 342: If the last byte of the normalized code stream generated by the previous double context symbol pair is a hexadecimal value OxFF, set the code value temporary left shift count value NewCT to 7, and then perform step 345; otherwise, execute Step 343;

Step 343: If the last byte of the normalized code stream generated by the previous double context symbol pair is a hexadecimal value OxFE, perform step 344; otherwise, set the code value temporary left shift count value NewCT to 8, and then execute Step 345;

Step 344: If the current code value C is greater than or equal to the hexadecimal value 0x8000000, set the code value temporary left shift count value NewCT to 7, otherwise set NewCT to 8; then perform step S345; step 345: code value C left Move NumSLA0-CT times, and then perform step 346;

Step 346: If the difference between the first probability interval value shift count value NumSLAO or the second probability interval value shift count value NumSLAl and the code value left shift count value CT is greater than or equal to 0, and is less than the code value temporary left shift count value NewCT, single byte output; otherwise two bytes output.

If the normalization type is a type of double normalization type, then a class of double normalization processing is performed. Referring to FIG. 12, the specific implementation is as follows:

Step S351: If NumSLA0-CT is less than zero, step S352 is performed; no shell I", step S353 is performed;

Step S352: If NumSLAl -(CT-NumSLA0) is less than zero, shift the code value C to the left by NumSLAO+NumSLAl times, and then end the current normalization process; otherwise, the shell lj, execute step S3521;

Step S3521: shift the code value C to the left by NumSLAO times, and update NewCT with NewCT-NumSLAO, and then perform step S3522;

Step S3522: If NumSLAl-NewCT is greater than or equal to zero, the code value C is left. Move NewCT times, and then perform step S3523; otherwise, end the current normalization process; Step S3523: If the last byte of the normalized code stream generated by the previous double context symbol pair is a hexadecimal value OxFF, set NewCT Is 7, then step S3526; otherwise, step S3524;

Step S3524: If the last byte of the normalized code stream generated by the previous double context symbol pair is a hexadecimal value OxFE, step S3525 is performed; otherwise, setting NewCT to 8, performing step S3526;

Step S3525: If the code value C is greater than or equal to the hexadecimal value 0x8000000, set NewCT to 7, otherwise, set NewCT to 8; then execute step S3526;

Step S3526: If the difference between NumSLAl and CT is smaller than NewCT, then a single byte output is performed, and then the current normalization process is ended; otherwise, two byte outputs are performed, and then the current normalization process is ended;

Step S353: shifting the code value C to the left by CT times, and executing step S3531;

Step S3531: If the last byte of the normalized code stream generated by the previous double context symbol pair is a hexadecimal value OxFF, set NewCT to 7, and execute step S3534; otherwise, execute step S3532: ;

Step S3532: If the last byte of the normalized code stream generated by the previous double context symbol pair is a hexadecimal value OxFE, perform step S3533; otherwise, set NewCT to 8, and perform step S3534;

Step S3533: If the code value C is greater than or equal to the hexadecimal value 0x8000000, set NewCT to 7, otherwise, set NewCT to 8; execute step S3534;

Step S3534: If the difference between the shift count NumSLAO and the CT is less than NewCT, step S3535 is performed, otherwise step S35310 is performed;

Step S3535: If the difference between the NumSLAl minus the CT is less than zero, the single byte output is performed, and then the current normalization process is ended; otherwise, step S3536 is performed;

Step S3536: If the last byte of the normalized code stream generated by the previous double context symbol pair is a hexadecimal value OxFF, set NewCT to 7, and then execute step S3539; otherwise, execute step S3537;

Step S3537: If the last byte of the normalized code stream generated by the previous double context symbol pair is a hexadecimal value OxFE, step S3538 is performed; otherwise, the NewCT is set to 8, Then step S3539 is performed;

Step S3538: If the code value C is greater than or equal to the hexadecimal value 0x8000000, set NewCT to 7, otherwise, set NewCT to 8; then execute step S3539;

Step S3539: If the difference between the shift count NumSLAl and the CT is smaller than NewCT, the byte output is performed twice, and then the current normalization process is ended; otherwise, the three-byte output is performed, and then the current normalization process is ended;

Step S35310: If the difference between the NumSLAl minus the CT is less than zero, the byte output is performed twice, and then the current normalization process is ended; otherwise, the step S35311 is performed;

Step S35311: If the last byte of the normalized code stream generated by the previous double context symbol pair is a hexadecimal value OxFF, set NewCT to 7, and then execute step S35314; otherwise, execute step S35312;

Step S35312: If the last byte of the normalized code stream generated by the previous double context symbol pair is a hexadecimal value of 0xFE, execute step S35313; otherwise, set NewCT to 8, and then execute step S35314;

Step S35313: If the code value C is greater than or equal to the hexadecimal value 0x8000000, set NewCT to 7, otherwise, set NewCT to 8; then execute step S35314;

Step S35314: If the difference between NumSLAl and CT is smaller than NewCT, the output is three bytes, and then the current normalization process is ended; otherwise, the fourth byte output is performed, and then the current normalization process is ended.

If the normalization type is a second-class double-normalization type, then the second-class double-order normalization processing is performed. Referring to FIG. 13, the specific implementation is as follows:

Step S361: If NumSLAO-CT is less than zero, step S362 is performed; otherwise, step S363 is performed;

Step S362: If NumSLAl-(CT-NumSLAO) is less than zero, first shift the code value C to the left by NumSLAO times, and add the left shifted code value to Qel, then shift the sum value to the left by NumSLAl times, and then end the current Normalization process; otherwise, step S3621 is performed;

Step S3621: First shift the code value C to the left by NumSLAO times, then update the code value C with the left-shifted code value and the Qel sum value, and update the NewCT with NewCT-NumSLAO, and execute step S3622;

Step S3622: If NumSLAl-NewCT is greater than or equal to zero, the code value C is left Shift NewCT times, and then step S3623; otherwise, end the current normalization process; Step S3623: If the last byte of the normalized code stream generated by the previous double context symbol pair is a hexadecimal value of 0xFF, set NewCT Is 7, then step S3626; otherwise, step S3624;

Step S3624: If the last byte of the normalized code stream generated by the previous double context symbol pair is a hexadecimal value OxFE, step S3625 is performed; otherwise, NewCT is set to 8, and then step S3626 is performed;

Step S3625: If the code value C is greater than or equal to the hexadecimal value 0x8000000, set NewCT to 7, otherwise, set NewCT to 8; then execute step S3626;

Step S3626: If NumSLAl-CT is smaller than NewCT, perform single byte output, and then end the current normalization process, otherwise perform two byte output, and then end the current normalization process;

Step S363: shift the code value C to the left by CT times, and then perform step S3631;

Step S3631: If the last byte of the normalized code stream generated by the previous double context symbol pair is a hexadecimal value of 0xFF, set NewCT to 7, and then execute step S3634; otherwise, execute step S3632;

Step S3632: If the last byte of the normalized code stream generated by the previous double context symbol pair is a hexadecimal value OxFE, execute step S3633; otherwise, set NewCT to 8, and then execute step S3634;

Step S3633: If the code value C is greater than or equal to the hexadecimal value 0x8000000, set NewCT to 7, otherwise, set NewCT to 8; then execute step S3634;

Step S3634: If the difference between the NumSLAO and the CT is smaller than the NewCT, first update the code value C with the sum of the code values C and Qel, and then perform step S3635; otherwise, first update the code value with the sum of the code value C and Qel. Then step S36310 is performed;

Step S3635: If NumSLAl-CT is less than zero, a single byte output is performed, and then the current normalization process is ended; otherwise, step S3636 is performed;

Step S3636: If the last byte of the normalized code stream generated by the previous double context symbol pair is a hexadecimal value of 0xFF, set NewCT to 7, and then execute step S3639; otherwise, execute step S3637;

Step S3637: If the last double context symbol pair is the last of the normalized code stream generated One byte is a hexadecimal value OxFE, step S3638 is performed; otherwise, NewCT is set to 8, and then step S3639 is performed;

Step S3638: If the code value C is greater than or equal to the hexadecimal value 0x8000000, set NewCT to 7, otherwise, set NewCT to 8; then execute step S3639;

Step S3639: If the difference between NumSLAl and CT is smaller than NewCT, perform two bytes of output, and then end the current normalization process; otherwise, perform three-byte output, and then end the current normalization process;

Step S36310: If NumSLAl-CT is less than zero, perform two byte output, and then end the current normalization process; otherwise, execute step S36311;

Step S36311: If the last byte of the normalized code stream generated by the previous double context symbol pair is a hexadecimal value OxFF, set NewCT to 7, and then execute step S36314; otherwise, execute step S36312;

Step S36312: If the last byte of the normalized code stream generated by the previous double context symbol pair is a hexadecimal value OxFE, step S36313 is performed; otherwise, NewCT is set to 8, and then step S36314 is performed;

Step S36313: If the code value C is greater than or equal to the hexadecimal value 0x8000000, set NewCT to 7, otherwise, set NewCT to 8; then execute step S36314;

Step S36314: If the difference between NumSLAl and CT is smaller than NewCT, the output is three bytes, and then the current normalization process is ended; otherwise, the fourth byte output is performed, and then the current normalization process is ended.

In Figure 8, Figure 9, and Figure 10, SHIFT_V(X) indicates that the operand X is shifted to the left until the X value after the left shift is greater than or equal to the hexadecimal value 0x8000.

In Figures 11, 12 and 13, the "previous code stream byte" represents the last byte of the normalized code stream produced by the last double context symbol pair.

In step S4, the code value is drained.

When there is no new context symbol pair input, the code value is drained and the corresponding empty code stream is obtained.

(classified cache output stream)

Step S5: classifying the buffer output code stream.

According to the number of output bytes of the code stream, the obtained single normalized code stream, one type of double normalized code The flow, the second-class double-normalized code stream, and the empty code stream are respectively classified and cached. The specific process is shown in Figure 14:

Step S51: If the code stream outputs only one byte, write the byte into the first byte buffer to prepare output.

Step S52: If the code stream outputs two bytes, the first byte is written into the first byte cache, and the second byte is written into the second byte buffer.

Step S53: If the code stream outputs three bytes, the first byte is written into the first byte buffer, the second byte is written into the second byte buffer, and the third byte is written. Into the third byte cache;

Step S54: If the code stream outputs four bytes, the first byte is written into the first byte buffer, the second byte is written into the second byte buffer, and the third byte is written. Enter the third byte buffer and write the fourth byte to the fourth byte buffer.

(output stream)

Step S6: Output the code stream of the classification cache.

Refer to Figure 15. The specific implementation of this step is as follows:

According to the sequential query manner, the code streams in the first to fourth byte buffers are sequentially read and output, and the arithmetic entropy coding of the dual context symbol pairs is completed. The specific process includes:

Step S61: If the first byte buffer is not empty, reading the data of one byte in the first byte buffer and outputting, and then performing step S62; otherwise, directly performing step S62;

Step S62: If the second byte buffer is not empty, reading the data of one byte in the second byte buffer and outputting, and then executing step S63; otherwise, directly executing step S63;

Step S63: If the third byte buffer is not empty, reading the data of one byte in the third byte buffer and outputting, and then executing step S64; otherwise, directly executing step S64;

Step S64: If the fourth byte buffer is not empty, the data of one byte in the fourth byte buffer is read and output, and then step S61 is performed; otherwise, step S65 is directly performed;

Step S65: If all the input context symbol pairs have been processed, the entire encoding process is ended, otherwise, the process returns directly to step S61.

The specific embodiments of the arithmetic entropy coding method of the present invention have been described in detail above. The effects of the present invention can be further illustrated by the following experimental data.

Table 4 shows the FPGA chip XC2V3000 and Altera using Xilinx respectively. The main technical specifications of the encoding method implemented by the FPGA chip STRATX include the FPGA resource utilization, ie the number of Slice and LC usage, the number of internally occupied memory bits, and the clock frequency and throughput rate associated with the processing speed.

Table 4 Main technical indicators of the coding system implemented by the present invention

As can be seen from Table 4, the present invention achieves arithmetic entropy coding of dual context symbol pairs in a single clock, with a significant increase in coding speed and efficiency.

Table 5 gives a qualitative comparison of the present invention with the prior art in terms of coding speed and implementation complexity. It can be seen that, in terms of processing speed, the present invention and the existing fifth encoding method have the highest processing speed, both of which reach 2 context/clock, but in terms of complexity, the current complexity of the fifth encoding method Significantly higher than the present invention, the internal storage capacity of the method is 8192 bits, while the internal storage capacity of the present invention is only 1509 bits.

Table 5 Comparison of the encoding speed and implementation complexity between the present invention and the prior art

As can be seen from Table 5, the present invention significantly improves the encoding speed, enables high-speed real-time encoding processing, and achieves low complexity.

The above description is only one specific embodiment of the present invention, and it is obvious to those skilled in the art that the present invention may be carried out in the form and details without departing from the principles and the structure of the present invention. Various modifications and changes are possible, but are intended to be included within the scope of the appended claims.

Claims

Claim

A high-speed real-time processing arithmetic entropy coding method based on the JPEG2000 standard, which is characterized in that:

The probability interval value determining step determines the coding symbol type according to the two context symbol pairs input in parallel, and updates the probability interval value and the coding parameter according to the coding symbol type;

a code value normalization step, updating the code value according to the above coded symbol type, determining the current normalization type according to the updated probability interval value and the coding parameter, and performing normalization processing, and outputting the normalized code stream And emptying the stream;

a code stream buffering step, according to the sequence of outputting the code stream, classifying and buffering the normalized code stream and the emptying code stream according to the outputted byte number of the normalized code stream and the emptying code stream, respectively, and concurrently Output;

In the outputting step, the normalized code stream and the empty code stream outputted in parallel in the code stream buffering step are sequentially serially output in a predetermined priority order.

2. The JPEG2000 standard-based high speed real-time processing arithmetic entropy coding method according to claim 1, wherein:

In the above probability interval value determining step, the method includes:

Steps of reading a double-coded index and a double-large probability symbol corresponding to the double context label, respectively, from two context symbol pairs consisting of a context label and a context decision entered in parallel;

Determining, according to the double-high probability symbol identifier and the dual context decision corresponding thereto, a probability type of a double context symbol pair, the probability type being a double probability symbol, a large probability/small probability symbol, a small probability, a large probability symbol, and a double Small probability symbol;

Determining the similarities and differences of the context labels according to the input double context label;

And step of jointly determining the type of the double context symbol pair according to the similarity and difference of the context label and the probability type of the double context symbol pair;

The step of updating the probability interval value according to the type of the double context symbol pair.

3. The JPEG2000 standard-based high-speed real-time processing arithmetic entropy coding method according to claim 2, wherein: The foregoing step of updating the probability interval value includes the following steps:

Obtaining, according to the obtained double-encoded index, a corresponding first small probability symbol probability value QeO and a second small probability symbol probability value Qel from the preset probability estimation table;

The first small probability symbol probability value QeO and the second small probability symbol probability value Qel are respectively saved into the first temporary variable temp2 and the second temporary variable tempO, and the current probability interval value A and the first small probability symbol probability value are simultaneously calculated. The difference of QeO Α-QeO, and save it to the third temporary variable tempi;

Updating the first and third temporary variable values respectively;

Updating the second temporary variable value according to the type of the double context symbol pair;

According to the relationship between the current probability interval value A and the first small probability symbol probability value QeO, the probability interval temporary variable value A_temp is respectively calculated for the double context symbol pairs of different probability types;

Correct the probability interval temporary variable value A_temp. If the probability interval temporary variable value A_temp is smaller than the hexadecimal value 0x8000, shift its value to the left by one bit, and then repeat the judgment process until A_temp is greater than or equal to ten. Hexadecimal value 0x8000;

The probability interval value A is updated to the corrected probability interval temporary variable value A_temp.

4. The JPEG2000 standard-based high speed real-time processing arithmetic entropy coding method according to claim 3, wherein:

Updating the first and third temporary variable values means updating by the relationship between the first and third temporary variable values and the hexadecimal value 0x8000, if the temporary variable value is less than the hexadecimal value 0x8000, The value of the temporary variable is shifted to the left by one bit, and the decision process is repeated thereafter until the temporary variable value is greater than or equal to the hexadecimal value 0x8000.

5. The JPEG2000 standard-based high speed real-time processing arithmetic entropy coding method according to claim 3, wherein:

Updating the second temporary variable value is done as follows:

If the type of the double context symbol pair is the same as the double large probability symbol or the large probability/small probability symbol is the same, the second temporary variable value is updated to the probability value that the next context symbol in the probability valuation table is a large probability symbol;

If the type of the double context symbol pair is the same as the double small probability symbol or the small probability / large probability symbol Similarly, the second temporary variable value is updated to a probability value in which the next context symbol in the probability estimate table is a small probability symbol.

6. The JPEG2000 standard-based high speed real-time processing arithmetic entropy coding method according to claim 3, wherein:

The above probability valuation table includes the following four sub-tables of different capacities:

(a) A small probability symbol probability value table having a depth of 32, a width of 13 bits, and a capacity of 32 x 13 bits, that is, 13 bits are used to represent the probability value;

(b) The next context symbol is a probability value table of a large probability symbol having a depth of 32, a width of 13 bits, and a capacity of 32 x 13 bits;

(c) The next context symbol is a probability value table of small probability symbols having a depth of 30, a width of 13 bits, and a capacity of 30 x 13 bits;

(d) A leading zero count table with a depth of 32, a width of 4 bits, and a capacity of 32x4 bits. Each leading zero count value stored in the table ranges from 1 to 15.

7. The JPEG2000 standard-based high speed real-time processing arithmetic entropy coding method according to claim 2, wherein:

In the above normalization step, it includes:

The steps of reading the probability interval value and the encoding parameter outputted in the above-described probability interval value determining step;

And updating the code value according to the normalization process identifier in the probability type and the coding parameter;

Determining a current normalization type according to the probability type, the probability interval value, the normalization process identifier, and the small probability symbol probability value;

Determining, according to the normalization type described above, a byte output type of the normalized code stream; and outputting the corresponding code stream byte according to the byte output type and the updated code value.

8. The JPEG2000 standard-based high speed real-time processing arithmetic entropy coding method according to claim 7, wherein:

In the case where the probability type of the pair of double context symbols is a double probability symbol, If the probability interval value is less than twice the first small probability symbol probability value, adjusting the first small probability symbol probability value; if the adjusted first small probability symbol probability value is less than twice the second small probability symbol probability value, The current normalization type is a type of double normalization, otherwise, the difference between the adjusted first small probability symbol probability value and the second small probability symbol probability value is calculated, if the difference is less than hexadecimal If the value is 0x8000, the current normalization type is a second-class double normalization, otherwise it is a single normalization; if the probability interval value is greater than or equal to twice the probability value of the first small probability symbol, the probability interval value is calculated. The difference between the probability values of a small probability symbol, and comparing the difference with the hexadecimal value 0x8000: If the difference is less than the hexadecimal value 0x8000, the probability interval value and the first small probability symbol probability value are adjusted The difference is: if the adjusted difference is less than twice the probability value of the second small probability symbol, the current normalization type is a type of double normalization, otherwise, the adjusted difference is compared with the second small Probability symbol probability value If the difference between the adjusted difference and the second small probability symbol probability value is less than the hexadecimal value 0x8000, the current normalization type is a second-class double normalization, otherwise a single normalization If the difference is greater than or equal to the hexadecimal value 0x8000, the relationship between the difference between the probability interval value and the first small probability symbol probability value and the second small probability symbol probability value: if the probability interval value is the first small If the difference between the probability symbol probability values is less than twice the second small probability symbol probability value, then the current normalization type is a single normalization, otherwise the difference between the probability interval value and the first small probability symbol probability value is calculated. The difference is less than the sum of the second small probability symbol probability value and the hexadecimal value 0x8000, then the current normalization type is a single normalization, otherwise zero normalization.

9. The JPEG2000 standard-based high speed real-time processing arithmetic entropy coding method according to claim 7, wherein:

In the case where the probability type of the pair of double context symbols is a large probability/small probability symbol,

If the probability interval value is less than twice the first small probability symbol probability value, adjusting the first small probability symbol probability value, if the adjusted first small probability symbol probability value is less than twice the second small probability symbol probability value, The current normalization type is the second-class double-order normalization, otherwise it is a class of double-order normalization; if the probability interval value is greater than or equal to twice the probability value of the first small probability symbol, and the probability interval value is the first small If the difference between the probabilistic symbol probability values is greater than or equal to the hexadecimal value 0x8000, the current normalization type is a single normalization. Otherwise, the difference between the probability interval value and the first small probability symbol probability value is adjusted. The subsequent difference is greater than or equal to twice the probability value of the second small probability symbol, then the current normalization type is a type of double normalization, otherwise it is a second type of double normalization.

10. The JPEG2000 standard-based high speed real time processing arithmetic entropy coding method according to claim 7, wherein:

In the case where the probability type of the pair of double context symbols is a small probability/high probability symbol,

If the probability interval value is greater than or equal to twice the first small probability symbol probability value, adjusting the first small probability symbol probability value, if the adjusted first small probability symbol probability value is less than twice the second small probability symbol probability value , the current normalization type is a second-class double-order normalization, otherwise, the difference between the adjusted first small probability symbol probability value and the second small probability symbol probability value is calculated, if the difference is less than sixteen If the value is 0x8000, the current normalization type is the second-class double normalization, otherwise it is a single normalization;

If the probability interval value is less than twice the first small probability symbol probability value, adjusting the difference between the probability interval value and the first small probability symbol probability value, if the adjusted difference is less than the second small probability symbol probability value If the current normalized type is a type of double normalization, otherwise, the adjusted difference and the second small probability symbol probability value are compared, if the adjusted difference and the second small probability symbol probability value The difference between the values is less than the hexadecimal value 0x8000, then the current normalization type is the second-class double-normalization, otherwise it is a single normalization.

11. The JPEG2000 standard-based high speed real-time processing arithmetic entropy coding method according to claim 7, wherein:

In the case where the probability type of the pair of double context symbols is a double small probability symbol, the normalized type is judged as a type of double normalization.

12. The JPEG2000 standard-based high speed real time processing arithmetic entropy coding method according to claim 7, wherein:

Updating the code value means that the sum of the second small probability symbol probability value and the current code value is calculated under the condition that the first context symbol pair is a large probability symbol, and the code value is updated to the sum value.

The high-speed real-time processing arithmetic entropy coding method based on the JPEG2000 standard according to any one of claims 8 to 10, characterized in that

Adjusting the first small probability symbol probability value means shifting the first small probability symbol probability value to the left several times, Until the value is greater than or equal to the hexadecimal value 0x8000.

The difference between the adjusted probability interval value and the first small probability symbol probability value is that the difference between the probability interval value and the first small probability symbol probability value is shifted left a number of times until the value is greater than or equal to the hexadecimal value 0x8000.

A high-speed real-time arithmetic arithmetic entropy coding method based on the JPEG2000 standard according to claim 7, wherein:

The high-speed real-time arithmetic arithmetic entropy coding method based on the JPEG2000 standard according to any one of claims 8 to 11, wherein

When classifying and buffering a single normalized code stream, a type of double normalized code stream, a second type of double normalized code stream, and an empty code stream, the following rules are performed:

(1) For a single-byte output stream, cache it in the first byte buffer;

(2) for the two-byte output code stream, it is sequentially buffered into the first and second byte buffers;

(3) for the three-byte output code stream, it is sequentially buffered into the first to third byte buffers;

(4) For the four-byte output code stream, it is sequentially buffered into the first to fourth byte buffers.