JP2010016684A - Encoding apparatus, encoding method and encoding program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To encode data utilizing solutions that have been discarded as an error in solutions obtained by solving an optimization problem using a principal algorithm. <P>SOLUTION: When an output from the principal algorithm (a solution) is inputted, a sweep out part identifies a row index (IndX) in a parity inspection matrix corresponding to an indefinite solution part that does not satisfy a parity inspection condition among outputs from the principal algorithm (solutions) using a sweep out method and stores it in a predetermined storage part. For each of indefinite solutions corresponding to the indefinite solution part, a quantizing part calculates a quantization value through a quantization operation performed with predetermined criteria. For indexes in the parity inspection matrix corresponding to the indefinite solution part, a solving part substitutes the respective quantization values calculated by the quantizing part to k simultaneous linear equations with n variables representing the parity inspection condition to result in the k simultaneous linear equations with k variables, solves the k simultaneous linear equations with the k variables, and calculates a solution x<SP>n</SP>that satisfies the parity inspection condition. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an encoding device, an encoding method, and an encoding program.

  Conventionally, in the field of communication technology, characters, images, voice, and other data are efficiently compressed with a small number of bits to store or transfer large amounts of data. There is an encoding technique for encoding data such as speech (see Non-Patent Document 1, for example).

In the encoding technique described above, for example, an optimization problem represented by the following expression (1) (hereinafter referred to as “optimization problem” as appropriate) is solved on a finite field F ₂ = {0, 1}. In general, a solution satisfying a predetermined constraint condition is selected as an optimum value from the obtained solutions and used as a code for compressing data.

  However, in order to solve the optimization problem strictly, an exponential order calculation amount is required, which is not efficient. Therefore, algorithms such as the sum-product method and the LP relaxation method (hereinafter referred to as “existing algorithm” as appropriate). A method for approximating the optimization problem using the above is proposed.

Note that “X ⁿ (= X _1,..., X _n )” in the above (1) is a random variable that appears in the same independent distribution according to the probability distribution P _X , and “A” is n rows. It is a matrix of k columns, and “c” is a horizontal vector of 1 row and k columns.

Tadashi Wadayama, “Low-density parity check code and its decoding method”, Trikes, 2002 ”

  However, among the solutions obtained by solving the optimization problem using the existing algorithm, a solution that does not satisfy the predetermined constraint condition (parity check condition) is processed as an error and discarded. It was.

  That is, some solutions that are processed as errors and are not used are not far from an optimum value that satisfies a predetermined constraint condition (parity check condition), and are valuable.

  Therefore, there is a need for a method of deriving a solution that satisfies a predetermined constraint condition using a solution that is processed as an error and is not used.

  Therefore, the present invention has been made to solve the above-described problems of the prior art, and has been processed as an error and discarded among solutions obtained by solving an optimization problem using an existing algorithm. Another object of the present invention is to provide an encoding device, an encoding method, and an encoding program capable of encoding data by utilizing the solution.

  In order to solve the above-described problems and achieve the object, the present invention is an encoding device that encodes data, and the encoding target data is stored in the storage unit that stores the encoding target data to be encoded. And encoding means for encoding the encoding target data using a predetermined algorithm, and an encoding result obtained by encoding the encoding target data by the encoding means satisfies a predetermined constraint condition A constraint condition determining unit that determines whether or not the encoding result is determined not to satisfy the predetermined constraint condition when the encoding result is determined not to satisfy the predetermined constraint condition. An indeterminate part deriving means for deriving an indeterminate part in the conversion result from the predetermined constraint condition, and each of the element values constituting the indeterminate part derived by the indeterminate part deriving means Quantized value calculating means for calculating each quantized quantized value, each quantized value calculated by the quantized value calculating means and each element value constituting the indeterminate portion are exchanged, and the predetermined value And an appropriate value calculating means for calculating an appropriate value that satisfies the constraint condition (1).

  Further, the present invention is an encoding method for encoding data, and reads the encoding target data from a storage unit that stores the encoding target data to be encoded, and sets the encoding target data to a predetermined value. An encoding step for encoding using the algorithm of: a constraint condition determining step for determining whether or not an encoding result obtained by encoding the data to be encoded by the encoding step satisfies a predetermined constraint condition; When the encoding result is determined not to satisfy the predetermined constraint condition in the constraint condition determination step, an indefinite portion in the encoding result that does not satisfy the predetermined constraint condition is determined as the predetermined constraint. Quantifying on a predetermined basis for each of the indeterminate part derivation step derived from the condition and each element value constituting the indeterminate part derived by the indeterminate part derivation step A quantized value calculating step for calculating each quantized value, and each quantized value calculated in the quantized value calculating step and each output value constituting the indeterminate portion are replaced to satisfy the predetermined constraint condition And a fitness calculation step for calculating a fitness value.

  Further, the present invention is an encoding program for causing a computer to execute a process for encoding data, reading the encoding target data from a storage unit that stores the encoding target data to be encoded, and An encoding procedure for encoding data to be encoded using a predetermined algorithm, and determining whether or not an encoding result obtained by encoding the data to be encoded by the encoding procedure satisfies a predetermined constraint condition If the encoding result is determined not to satisfy the predetermined constraint condition by the constraint condition determination procedure to be performed and the constraint condition determination procedure, an indefinite in the encoding result that does not satisfy the predetermined constraint condition For each of the indeterminate part derivation procedure for deriving a part from the predetermined constraint condition and the element values constituting the indeterminate part derived by the indeterminate part derivation procedure Quantized value calculation procedures for calculating quasi-quantized quantized values respectively, and replacing each quantized value calculated by the quantized value calculating procedure with each element value constituting the indeterminate portion, And a fitness value calculation procedure for calculating a fitness value that satisfies the constraint conditions.

  Advantageous Effects of Invention According to the present invention, there is an effect that data can be encoded using a solution that has been processed as an error and discarded, among solutions obtained by solving an optimization problem using an existing algorithm. .

  Exemplary embodiments of an encoding device, an encoding method, and an encoding program according to the present invention will be described below in detail with reference to the accompanying drawings. In the following, the outline of the encoding device according to the present invention, the configuration and processing of the encoding device according to the present invention will be described in order, and finally the effects of the first embodiment will be described.

[Explanation of terms]
The main terms used in Example 1 below will be described. “N × k matrix: A” in the common parameters appearing in the first embodiment below is a parity check matrix whose matrix elements are “0” or “1”, and constitutes a parity check condition.

  Further, “1 × k vector: c” in the common parameters appearing in the first embodiment below is a constant vector whose elements are “0” or “1”, and is arbitrarily given in advance.

In addition, “probability distribution: P _X ” appearing as a parameter in Example 1 below represents a probability distribution on a finite field F ₂ = {0, 1}. For example, “P _X ⁿ ” includes n pieces of probability distributions. Represents a probability distribution for an independent distribution series.

In addition, the output “z ⁿ ” from the main algorithm described in the first embodiment is an optimization problem for various data (characters, images, sounds, etc.) stored in the information source (see the above formula (1)). approximately solved resulting solution to a (reference), the output "z ^n" elements z _i constituting the (
1 ≦ i ≦ n) satisfies 0 ≦ z _i ≦ 1.

In addition, “q (z ⁿ ) A = c” (or “x ⁿ A = c”: k simultaneous linear equations with n variables) appearing in Example 1 below is an output “z ⁿ ” from the main algorithm. Is a parity check condition to be satisfied (constraint condition of the optimization problem shown in the above (1)), and “q (z ⁿ )” is each element z _i (which constitutes the output “z ⁿ ” from the main algorithm). 1 ≦ i ≦ n) is quantized to “0” or “1” according to a predetermined standard (depending on the main algorithm).

In addition, “x ⁿ ” appearing in the following first embodiment represents the final output of the encoding apparatus according to the first embodiment, and includes “0” or “1” elements.

[Outline of Encoding Device (Example 1)]
First, the gist of the encoding apparatus according to the first embodiment will be described with reference to FIGS. 1 and 2. FIG. 1 is a diagram illustrating an overall image of the algorithm of the encoding device according to the first embodiment. FIG. 2 is a diagram illustrating the configuration of the auxiliary algorithm according to the first embodiment.

  The encoding apparatus according to the first embodiment uses the main algorithm (existing algorithm) for the purpose of efficiently compressing various data (characters, images, sounds, etc.) stored in the information source. The outline is to encode various data with the solution obtained by solving the above.

  The essence of the encoding apparatus according to the first embodiment is to encode data using a solution that is processed as an error and thrown away among solutions obtained by solving an optimization problem using the main algorithm. There is a point that can be done.

  Specifically, as shown in FIG. 1, the encoding apparatus according to the first embodiment is obtained by a main algorithm that approximately solves an optimization problem (see the above formula (1)) using each parameter. When the solution does not satisfy the parity check condition, the essence of the encoding apparatus according to the first embodiment is realized by an auxiliary algorithm that performs processing on the solution obtained by the main algorithm.

  Specifically, as shown in FIG. 2, the auxiliary algorithm of the encoding apparatus according to the first embodiment includes a sweeping unit, a quantizing unit, and a solution finding unit.

Then, in the encoding device according to the first embodiment, whether or not the output (solution) from the main algorithm satisfies the k simultaneous linear equations (“x ⁿ A = c”) of n variables representing the parity check condition. judge.

  In the encoding apparatus according to the first embodiment, when it is determined that the output (solution) from the main algorithm does not satisfy the parity check condition due to the determination result, for example, n> k. The output (solution) from the main algorithm is output to the auxiliary algorithm.

  Therefore, when the output from the main algorithm (solution) is input to the sweeping unit of the auxiliary algorithm, in the parity check matrix corresponding to the indefinite solution part that does not satisfy the parity check condition in the output (solution) from the main algorithm. The row index (IndX) is specified using the sweep method and stored in a predetermined storage unit (see (1) in FIG. 2).

  Next, the quantization unit of the auxiliary algorithm performs a quantization operation performed on a predetermined basis for each indefinite solution corresponding to the indefinite solution part that does not satisfy the parity check condition in the output (solution) from the main algorithm. The quantized values are respectively calculated by (see (2) in FIG. 2).

Subsequently, the solution calculation unit of the auxiliary algorithm uses the quantization value calculated by the quantization unit for the index corresponding to the indefinite solution part that does not satisfy the parity check condition in the output (solution) from the main algorithm. Are substituted into k simultaneous linear equations of n variables representing parity check conditions, respectively, and are reduced to k simultaneous linear equations of k variables, and a solution x ⁿ satisfying the parity check conditions is solved by solving the k simultaneous linear simultaneous equations of k variables. Is calculated (see (3) in FIG. 2).

The encoding apparatus according to the first embodiment encodes data with the solution ^xn calculated by the solution finding unit.

  For this reason, the encoding apparatus according to the first embodiment uses the solution that has been processed as an error and discarded, among the solutions obtained by solving the optimization problem using the main algorithm. Can be encoded.

[Configuration of Encoding Device (Example 1)]
Next, the configuration of the encoding apparatus according to the first embodiment will be described with reference to FIGS. 3 and 4. FIG. 3 is a diagram illustrating an application example of the encoding apparatus according to the first embodiment. FIG. 4 is a diagram illustrating the configuration of the encoding device according to the first embodiment.

  As illustrated in FIG. 3, the encoding apparatus according to the first embodiment, for example, transmits encoded data obtained by encoding data (characters, images, and the like) stored in an information source by an encoding unit via a communication path. To the decoding device.

  In the decoding device that has received the encoded data from the encoding device, the decoding unit decodes the encoded data to acquire data.

  The encoding apparatus described above includes, for example, an information storage unit 100, an optimization execution unit 200, and an auxiliary encoding unit 300 as illustrated in FIG.

  The information storage unit 100 is a storage unit that stores various data such as characters, images, and sounds.

  In order to encode various data stored in the information storage unit 100, the optimization execution unit 200 uses an existing main algorithm such as a sum-product method or an LP relaxation method (for example, ( This is a processing unit that approximates 1).

The optimization execution unit 200 outputs an output z ⁿ that is a calculation result to the auxiliary encoding unit 300 described later.

Auxiliary coding part 300, the output z ⁿ input from the optimization execution unit 200, when not satisfied parity check condition is the processing unit that performs processing for output z ⁿ input from the optimization execution unit 200 .

  For example, as shown in FIG. 4, the auxiliary encoding unit 300 includes a matrix coefficient storage unit 310, an output determination unit 320, a sweep calculation processing unit 330, a row index storage unit 340, and a quantization calculation unit 350. And a solution calculation unit 360.

  The output determination unit 320 corresponds to the “constraint condition determination unit” described in the claims, and the sweep calculation processing unit 330 corresponds to the “undefined part derivation unit” described in the claims, The quantization operation unit 350 corresponds to the “quantized value calculation unit” described in the claims, and the solution calculation unit 360 corresponds to the “adaptation value calculation unit” described in the claims.

  The matrix coefficient storage unit 310 is a storage unit that stores matrix coefficients of the parity check matrix A.

The output determination unit 320 quantizes the output z ⁿ input from the optimization execution unit 200 to “0” or “1” according to a predetermined reference, and the quantized value is converted into a parity check condition (“x ⁿ A = C ").

Specifically, when the output determination unit 320 receives the output z ⁿ from the optimization execution unit 200, the output determination unit 320 reads the matrix coefficient of the parity check matrix A from the matrix coefficient storage unit 310 and sets the parity check condition (k variables of n variables). The linear equation: “x ⁿ A = c”) is calculated.

When the output determination unit 320 determines that the output z ⁿ input from the optimization execution unit 200 does not satisfy the parity check condition (k simultaneous linear equations of n variables: “x ⁿ A = c”). Outputs the output z ⁿ to the quantization operation unit 350 described later.

On the other hand, when the output determination unit 320 determines that the output z ⁿ input from the optimization execution unit 200 satisfies the parity check condition (“x ⁿ A = c”), the output z ⁿ is finally output. Output as ^xn .

In the following, for convenience, a case where the following condition (2) is adopted for the n-variable k simultaneous linear equations “x ⁿ A = c”, which is a parity check condition, will be described.

The sweep calculation processing unit 330 is determined by the output determination unit 320 that the output z ⁿ input from the optimization execution unit 200 does not satisfy the parity check condition (“x ⁿ A = c”). when, performing the parity check condition ( "x ^{n a} = c") sweep arithmetic processing for specifying the row index in the parity check condition corresponding to the undefined solution portion is not satisfied ( "x ^{n a} = c") the It is a processing unit.

  Specifically, the sweep calculation processing unit 330 reads matrix coefficients from the matrix coefficient storage unit 310 to generate, for example, the parity check matrix A shown in the above condition (2), and generates a column of the generated parity check matrix A Perform basic deformation. The parity check matrix after column basic modification is matrix (3) shown below.

Then, the sweep calculation processing unit 330 calculates the indexes of the remaining rows by excluding the row giving the k (= 3) -dimensional unit vector from the parity check matrix after the column basic deformation (see the matrix (3) above). It is specified as a row index (IndX) in the parity check condition corresponding to an indefinite solution part that does not satisfy the parity check condition (“x ⁿ A = c”), and is stored in the row index storage unit 340.

For example, when the above condition (2) is satisfied, the sweep calculation processing unit 330 sets IndX = {2, as an index (IndX) corresponding to an indefinite solution part that does not satisfy the parity check condition (“x ⁿ A = c”). 3} (second row, third row) is stored in the row index storage unit 340.

  The row index storage unit 340 is a storage unit that stores an indefinite solution part index (IndX). For example, IndX = {2, 3} stored from the sweep calculation processing unit 330 is a parity check corresponding to the indefinite solution part. Store as an index (IndX) within the condition.

The quantization operation unit 350 performs the quantization operation on the indefinite solution part that does not satisfy the parity check condition (“x ⁿ A = c”) in the output z ⁿ input from the optimization execution unit 200 according to a predetermined criterion. It is a processing unit that performs a quantization value calculation.

Specifically, when the output z ⁿ is input from the optimization execution unit 200, the quantization calculation unit 350 receives an indefinite solution portion (z _i ) of the output z ⁿ that does not satisfy the parity check condition (“x ⁿ A = c”). ) Is quantized according to the following criterion (4), a quantized value Q (z _i ) is calculated and output to the solution calculation unit 360.

The solution calculation unit 360 performs a quantization operation on the quantization value obtained by performing the quantization operation on the indefinite solution part (z _i ) of the output z ⁿ that does not satisfy the parity check condition (“x ⁿ A = c”). This is a processing unit that calculates a solution x ⁿ that is input from the unit 350 and satisfies the parity check condition (“x ⁿ A = c”).

More specifically, when the quantized value is input from the quantization calculating unit 350, the solution calculating unit 360 receives an indefinite solution based on the row index (IndX = {2, 3}) read from the row index storage unit 340. The quantized value Q (z _i ) is substituted for the value in the equation “x ⁿ A = c” representing the parity check condition corresponding to the portion, x _i (i∈IndX = {2, 3}).

Then, the solution calculation unit 360 assigns the quantized value Q (z _i ) to x _i (iεIndX = {2, 3}), and then develops the equation “x ⁿ A = c” representing the parity check condition. Then, a solution x ⁿ satisfying the parity check condition (“x ⁿ A = c”) is calculated by solving the k (= 3) simultaneous linear equations of the k (= 3) variable.

  Under the condition (2), the equation to be solved by the solution calculation unit 360 is as shown in the following equation (5).

[Processing of Encoding Device (Example 1)]
Subsequently, the processing of the encoding apparatus according to the first embodiment will be described with reference to FIG. FIG. 5 is a diagram illustrating a process flow of the encoding device according to the first embodiment.

As shown in the figure, the output determining unit 320 determines whether or not the output z ⁿ input from the optimization executing unit 200 satisfies the parity check condition (“x ⁿ A = c”) (step S501). ).

As a result of the determination, the output determination unit 320 determines that the output z ⁿ input from the optimization execution unit 200 does not satisfy the parity check condition (k simultaneous linear equations of n variables: “x ⁿ A = c”). In such a case (No at step S501), the output z ⁿ is output to the quantization calculation unit 350.

The sweep calculation processing unit 330 includes an indefinite solution portion that does not satisfy the parity check condition (“x ⁿ A = c”) in the output z ⁿ input from the optimization execution unit 200 by the output determination unit 320. If it is determined to be performed, a sweep calculation process for specifying an index in the parity check condition (“x ⁿ A = c”) corresponding to the indefinite solution portion is performed (step S502).

  Specifically, the sweep calculation processing unit 330 reads a matrix coefficient from the matrix coefficient storage unit 310, generates a parity check matrix A (see the above condition (2)), and generates a column base of the generated parity check matrix A Perform the transformation.

  Then, the sweep calculation processing unit 330 calculates the indexes of the remaining rows by excluding the row giving the k (= 3) -dimensional unit vector from the parity check matrix after the column basic deformation (see the matrix (3) above). The index (IndX) in the parity check condition corresponding to the indefinite solution part is specified and stored in the row index storage unit 340.

For example, when the above condition (2) is satisfied, the sweep calculation processing unit 330 sets IndX = {2, as an index (IndX) corresponding to an indefinite solution part that does not satisfy the parity check condition (“x ⁿ A = c”). 3} (second row, third row) is stored in the row index storage unit 340.

The quantization operation unit 350 performs the quantization operation on the indefinite solution part that does not satisfy the parity check condition (“x ⁿ A = c”) in the output z ⁿ input from the optimization execution unit 200 according to a predetermined criterion. The quantized value is calculated (step S503).

Specifically, when the output z ⁿ is input from the optimization execution unit 200, the quantization calculation unit 350 receives an indefinite solution portion (z _i ) of the output z ⁿ that does not satisfy the parity check condition (“x ⁿ A = c”). ) Is quantized according to a predetermined criterion (see the above criterion (4)), a quantized value Q (z _i ) is calculated and output to the solution calculating unit 360.

The solution calculation unit 360 performs a quantization operation on the quantization value obtained by performing the quantization operation on the indefinite solution part (z _i ) of the output z ⁿ that does not satisfy the parity check condition (“x ⁿ A = c”). The solution x ⁿ that is input from the unit 350 and satisfies the parity check condition (“x ⁿ A = c”) is calculated (step S504).

More specifically, when the quantized value is input from the quantization calculating unit 350, the solution calculating unit 360 receives an indefinite solution based on the row index (IndX = {2, 3}) read from the row index storage unit 340. The quantized value Q (z _i ) is substituted for the value in the equation “x ⁿ A = c” representing the parity check condition corresponding to the portion, x _i (i∈IndX = {2, 3}).

Then, the solution calculation unit 360 assigns the quantized value Q (z _i ) to x _i (iεIndX = {2, 3}), and then develops the equation “x ⁿ A = c” representing the parity check condition. Then, a solution x ⁿ satisfying the parity check condition (“x ⁿ A = c”) is calculated by solving the k (= 3) simultaneous linear equations of the k (= 3) variable.

The solution calculation unit 360 outputs the calculated solution ^xn as a final output (step S505), and the encoding apparatus ends the process.

Here, returning to the description of step S501, the output determination unit 320 determines that the output z ⁿ input from the optimization execution unit 200 as a result of the determination is a parity check condition (k simultaneous linear equations of n variables: “x ⁿ A = c ") if it is determined as to satisfy the (Yes at step S501), and outputs the output ^{z n} as the final output ^{x n} (step S506), the encoding device ends the process.

[Effects of Example 1]
As described above, according to the first embodiment, when the output (solution) from the optimization execution unit 200 does not satisfy the parity check condition, the output (solution) from the optimization execution unit 200 Since the auxiliary encoding unit 300 calculates a solution that satisfies the parity check condition using a value obtained by quantizing the indefinite solution part that does not satisfy the parity check condition, the solution that has been processed as an error and discarded is used. Thus, there is an effect that the data can be encoded.

  In addition, the processing of the auxiliary encoding unit 300 described in the first embodiment (see, for example, FIG. 5) is performed using a distortion information source encoding method (S.Miyake, J.Marumatsu, “Construction of lossy”) that is an existing technology. code using LDPC matrices, "Proc. 2007 IEEE Int. Symp. Inform Theory, pp. 1106-1110, 2007)).

  For example, in a distorted information source coding scheme, it is obtained by encoding information reduction using the LP relaxation method (see J. Feldman, Decoding Error-Correcting Codes via Linear Programming, PhD dissertation, Massachusetts Institute of Technology, 2003) By applying the processing of the auxiliary encoding unit 300 described in the first embodiment to the output in the same manner, a final output that meets the parity check condition can be obtained.

  Although the embodiments of the encoding apparatus of the present invention have been described so far, the present invention may be implemented in various different forms other than the above-described embodiments. Therefore, another embodiment included in the present invention will be described below.

(1) Device Configuration, etc. Each component of the encoding device shown in FIG. 4 is functionally conceptual and does not necessarily need to be physically configured as illustrated.

  That is, the specific form of distribution / integration of the encoding device shown in FIG. 4 is not limited to that shown in the figure. For example, the output determination unit 320 of the auxiliary encoding unit 300, the quantization calculation unit 350, and the solution calculation calculation unit 360 All or a part of it, such as a single processing function unit, can be configured functionally or physically distributed and integrated in arbitrary units according to various loads and usage conditions. it can.

  Furthermore, each processing function (see, for example, FIG. 5) performed by the encoding device shown in FIG. 4 is realized by a CPU and a program that is analyzed and executed by the CPU. Alternatively, it can be realized as hardware by wired logic.

(2) Encoding program Various processes of the encoding apparatus described in the first embodiment (for example, see FIG. 5 and the like) are executed by a computer system such as a personal computer or a workstation. Can be realized.

  In the following, an example of a computer that executes an encoding program having the same function as that of the encoding apparatus (auxiliary encoding unit 300) described in the first embodiment will be described with reference to FIG. FIG. 6 is a diagram illustrating a computer that executes an encoding program.

  As shown in the figure, a computer 400 as an encoding device (auxiliary encoding unit 300) is configured by connecting an input unit 410, an output unit 420, an HDD 430, a RAM 440 and a CPU 450 via a bus 500.

  Here, the input unit 410 receives input of various data from the user. The output unit 420 displays various information. The HDD 430 stores information necessary for the CPU 450 to execute various processes. The RAM 440 temporarily stores various information. The CPU 450 executes various arithmetic processes.

  As shown in FIG. 6, the HDD 430 includes an encoding program 431 that exhibits the same function as each processing unit of the encoding apparatus (auxiliary encoding unit 300) described in the first embodiment, and an encoding Processing data 432 is stored in advance.

  Note that the encoding program 431 may be appropriately distributed and stored in a storage unit of another computer that is communicably connected via a network.

  Then, the CPU 450 reads out the encoding program 431 from the HDD 430 and develops it in the RAM 440, so that the encoding program 431 functions as an encoding process 441 as shown in FIG.

  Then, the encoding process 441 reads the encoding process data 432 and the like from the HDD 430, expands them in the area allocated to itself in the RAM 440, and executes various processes based on the expanded data and the like.

  Note that the encoding process 441 is executed by the output determination unit 320, the sweep calculation processing unit 330, the quantization calculation unit 350, the solution calculation calculation unit 360, and the like of the auxiliary encoding unit 300 included in the encoding device illustrated in FIG. It corresponds to each processing.

  Note that the above-described encoding program 431 is not necessarily stored in the HDD 430 from the beginning. For example, a flexible disk (FD), a CD-ROM, a DVD disk, a magneto-optical disk, an IC inserted into the computer 400, and the like. Each program is stored in a “portable physical medium” such as a card, and “another computer (or server)” connected to the computer 400 via a public line, the Internet, a LAN, a WAN, or the like. The computer 400 may read and execute each program from these.

(3) Coding method The following coding method is realized by the coding apparatus (auxiliary coding unit 300) described in the first embodiment.

  That is, an encoding step of reading the encoding target data from a storage unit that stores the encoding target data to be encoded, and encoding the encoding target data using a predetermined algorithm, and an encoding step And a constraint condition determination step for determining whether or not an encoding result obtained by encoding the data to be encoded satisfies a predetermined constraint condition (see, for example, step S501 in FIG. 5), and does not satisfy the predetermined constraint condition When it is determined by the constraint condition determination step that an indeterminate part exists in the encoding result, an indeterminate part derivation step for deriving the indeterminate part from a predetermined constraint condition (see, for example, step S502 in FIG. 5) ), For each output value constituting the indeterminate part determined in the indeterminate part derivation step, calculate a quantized value quantized according to a predetermined criterion The quantized value calculation step (see, for example, step S503 in FIG. 5) is replaced with each quantized value calculated in the quantized value calculating step and each output value constituting the indeterminate portion to satisfy a predetermined constraint condition. An encoding method including an output value calculating step for calculating such an output value (see, for example, step S504 in FIG. 5) is realized.

  As described above, the encoding device, the encoding method, and the encoding program according to the present invention are for the purpose of efficiently compressing various data (characters, images, sounds, etc.) stored in an information source. It is useful for encoding various data with solutions obtained by solving optimization problems using the main algorithm (existing algorithm). Especially, among the solutions obtained by solving optimization problems using the main algorithm Therefore, it is suitable for encoding data using the solution that has been processed as an error and discarded.

It is a figure which shows the whole image of the algorithm of the encoding apparatus which concerns on Example 1. FIG. It is a figure which shows the structure of the auxiliary | assistant algorithm which concerns on Example 1. FIG. FIG. 3 is a diagram illustrating an application example of the encoding device according to the first embodiment. 1 is a diagram illustrating a configuration of an encoding apparatus according to a first embodiment. It is a figure which shows the flow of a process of the encoding apparatus which concerns on Example 1. FIG. It is a figure which shows the computer which performs an encoding program.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Information memory | storage part 200 Optimization execution part 300 Auxiliary encoding part 310 Matrix coefficient memory | storage part 320 Output determination part 330 Sweep arithmetic processing part 340 Row index memory | storage part 350 Quantization arithmetic part 360 Solving calculation part 400 Computer (encoding apparatus (auxiliary Encoding part))
410 Input unit 420 Output unit 430 HDD (Hard Disk Drive)
431 Encoding program 432 Encoding processing data 440 RAM (Random Access Memory)
441 Encoding process 450 CPU (Central Processing Unit)
500 buses

Claims (4)

An encoding device for encoding data,
An encoding unit that reads the encoding target data from a storage unit that stores the encoding target data to be encoded, and encodes the encoding target data using a predetermined algorithm;
Constraint condition determination means for determining whether or not an encoding result obtained by encoding the data to be encoded by the encoding means satisfies a predetermined constraint condition;
If it is determined by the constraint condition determination means that the encoding result does not satisfy the predetermined constraint condition, an indefinite portion in the encoding result that does not satisfy the predetermined constraint condition is determined as the predetermined constraint condition. An indeterminate part deriving means derived from
Quantized value calculating means for calculating a quantized value quantized according to a predetermined standard for each of the element values constituting the indeterminate part derived by the indeterminate part deriving means;
Replacing each quantized value calculated by the quantized value calculating means with each element value constituting the indeterminate portion, and adapted value calculating means for calculating an adapted value that satisfies the predetermined constraint condition;
An encoding device comprising: The encoding means outputs a solution that is an encoding result obtained by encoding the encoding target data using an algorithm that approximates maximum likelihood estimation to the constraint condition determination means,
The constraint condition determination means determines whether or not the solution input from the encoding means satisfies an equation representing the predetermined constraint condition,
The indeterminate part deriving unit, when the constraint condition determining unit determines that the solution input from the encoding unit does not satisfy the equation representing the predetermined constraint condition, a check matrix constituting the equation To derive a row index in the parity check matrix corresponding to the indefinite solution portion of the solution that does not satisfy the equation,
The quantized value calculating means calculates a quantized value obtained by quantizing a solution corresponding to the row index derived by the indeterminate portion deriving means according to a predetermined criterion;
The calculating means replaces each quantized value calculated by the quantized value calculating means with a solution corresponding to the row index, and calculates a solution satisfying an equation representing the predetermined constraint condition. The encoding device according to claim 1. An encoding method for encoding data, comprising:
An encoding step of reading the encoding target data from a storage unit that stores the encoding target data to be encoded, and encoding the encoding target data using a predetermined algorithm;
A constraint condition determining step for determining whether or not an encoding result obtained by encoding the data to be encoded in the encoding step satisfies a predetermined constraint condition;
If it is determined in the constraint condition determination step that the encoding result does not satisfy the predetermined constraint condition, an indefinite portion in the encoding result that does not satisfy the predetermined constraint condition is determined as the predetermined constraint condition. An indeterminate part derivation step derived from
A quantized value calculating step for calculating a quantized value quantized according to a predetermined criterion for each of the element values constituting the indeterminate portion derived by the indeterminate portion deriving step;
An adaptive value calculating step of calculating an adaptive value that satisfies the predetermined constraint condition by replacing each quantized value calculated in the quantized value calculating step with each output value constituting the indeterminate portion;
The encoding method characterized by including. An encoding program for causing a computer to execute a process for encoding data,
An encoding procedure for reading the encoding target data from a storage unit that stores the encoding target data to be encoded, and encoding the encoding target data using a predetermined algorithm;
A constraint condition determination procedure for determining whether an encoding result obtained by encoding the encoding target data by the encoding procedure satisfies a predetermined constraint condition;
When it is determined by the constraint condition determination procedure that the encoding result does not satisfy the predetermined constraint condition, an indefinite portion in the encoding result that does not satisfy the predetermined constraint condition is determined as the predetermined constraint condition. An indeterminate part derivation procedure derived from
A quantized value calculation procedure for calculating a quantized value quantized according to a predetermined standard for each of the element values constituting the indeterminate portion derived by the indeterminate portion derivation procedure;
An adaptive value calculation procedure for calculating an adaptive value that satisfies the predetermined constraint condition by replacing each quantized value calculated by the quantized value calculation procedure with each element value constituting the indeterminate part;
An encoding program for causing a computer to execute.

Images