WO1998054841A1

WO1998054841A1 - Data compressing device by permutation encoding and decompressing device

Info

Publication number: WO1998054841A1
Application number: PCT/JP1997/001860
Authority: WO
Inventors: Zhongshu Ji
Original assignee: Kubota Systems Inc.
Priority date: 1997-05-30
Filing date: 1997-05-30
Publication date: 1998-12-03
Also published as: AU2977497A

Abstract

A device and method for compressing data by which data can be compressed quickly at high compressibility. Image information given to a two-dimensional DCT means (2) is converted into two-dimensional DCT converted data. The two-dimensional DCT data thus obtained are scanned in zigzag from the lowest frequency component by means of a zigzag-scanning means (4). An aligning means (6) roughly sorts the DCT data in order of amplitude from the largest one in accordance with prescribed roughness. Therefore, sorted alignment value information and data order changing information which is produced as a result of the sorting are obtained. Inverse permutation information is used as the data order changing information. The alignment value information is encoded by means of an alignment value encoding means (10) and outputted as encoded alignment value information. The inverse permutation information obtained by means of the means (6) is given to a PCRing means (14) and the PCR image of the information is calculated. The PCR image is encoded into Hafman codes by means of a Hafman-encoding means (18) and outputted as an encoded PCR image. An output means (20) outputs the above-mentioned encoded alignment value information and encoded PCR image as compressed image data.

Description

Description Data compression device and decompression device by permutation coding

The present invention relates to data compression, and more particularly, to improving the compression ratio. Background art

The inventor has already proposed a new compression method based on the permutation group theory. (Kinaka Shu et al., "A row permutation coding method for images characterized by high speed and high image quality." No. 2, 1991: Zhongshu Ji et al "Block Permutation Coding of Images Using Cosine Transform" IEEE Transactions on Communica Hons Vol 43, Nol 1, 1995). It has been verified by the inventor that the compression ratio can be improved by using the compression method based on the permutation group theory as compared with the commonly used JPEG. However, further improvements in compression ratio and compression processing speed are desired. Summary of the Invention

SUMMARY OF THE INVENTION An object of the present invention is to provide a data compression logic and a coding technique to which permutation group theory with improved performance is applied.

The data compression device of claim 1 is

Receiving a plurality of data given in an order, outputs the value indicated by each data according to a predetermined rule, outputs sorted value information, and compares the sorted order obtained by the above sorting with the given order. Alignment means that outputs the following reverse replacement information indicating the correspondence,

= (<1) Yu (2) Yu (3) Φ duck

Here, is m that satisfies σ (πι) in the following replacement information σ, σ = (σ (1) σ (2) σ (3) σ (Μ))

Here, σφ indicates the number of the j-th data after sorting in the order before sorting, and M indicates the number of data.

PCR conversion means for receiving the inverse permutation information obtained by the alignment means and outputting the following PCR map m for the inverse permutation information Φ:

Here, the k-th element m _k is obtained by the following equation:, Va (k) <h (k + 1)

OK

k-nk-1, Va (k)> hi (k + 1)

k = 2, 3, 4,, M where, m _k is the (Kooite, 0 (sigma (1) or k matrix elements and <Ha (k +1)) i.e. k + 1 matrix elements Is the number of elements whose value is less than k,

PCR encoding means for encoding the PCR map m obtained by the PCR encoding means and outputting an encoded PCR map,

An alignment value encoding unit that encodes the alignment value information obtained by the alignment unit and outputs the encoded alignment value information;

Output means for outputting the encoded PCR mapping from the PCR encoding means and the encoded alignment value information from the alignment value encoding means as compressed data.

The data compression apparatus according to claim 2 is characterized by further comprising orthogonal transformation means for orthogonally transforming the given original data and providing the data as alignment target data to the alignment means.

The data compression device according to claim 3 is characterized in that the orthogonal transformation means receives the original data, performs DCT transformation, obtains zigzag-scanned DCT transformation data, and gives it to the sorting means as sorting target data. I have.

The data compression apparatus according to claim 4 receives the DCT transform data, quantizes the DCT transform data based on a predetermined quantization roughness, and the lowest of the quantized DCT transform data. From the frequency components, those with the highest frequency components exceeding a predetermined value

-2-Replacement form (Rule 26) And an effective coefficient extracting means for extracting the effective coefficients to the sorting means as sorting target data.

The data compression apparatus according to claim 5 is characterized in that the PCR encoding means encodes the PCR map into Huffman encoding.

The data compression apparatus according to claim 6 is characterized in that the alignment value encoding means performs encoding by expressing the alignment value as a function.

A data compression device according to claim 7, wherein the alignment value encoding means performs encoding by expressing a value obtained by multiplying the alignment value by an expansion coefficient as a function.

The data compression method of claim 8 is

Receives a plurality of data given in order, obtains sorted value information in which the values indicated by each data are sorted according to a predetermined rule,

In order to show the correspondence between the sorted order obtained by the above sorting and the given order, order information in which the sorted order is arranged in accordance with the given order is obtained, and the order information is arranged in a permutation. Obtaining structured order mapping information,

It is characterized in that each of the alignment value information and the order mapping information is encoded to obtain compressed data.

The data decompression device of claim 9 is

PCR decoding means for receiving the encoded PCR map in the given compressed data, decoding the encoded PCR map, and obtaining the PCR map;

Inverse PCR conversion means for receiving the PCR mapping obtained by the PCR decoding means and obtaining reverse replacement information corresponding to the PCR mapping,

An alignment value decoding means for receiving the encoding alignment value information in the given compressed data, decoding the encoding alignment value information, and obtaining the alignment value information;

Restoring means for restoring original data based on the reverse replacement information obtained by the inverse PCR converting means and the alignment value information obtained by the alignment value decoding means;

It has.

The data decompression method of claim 10 is

By decomposing the given compressed data, the coding PCR map and the coding alignment value information are separated. Get,

Receiving the coded PCR map obtained by the above decomposition, decoding the coded PCR map to obtain a PCR map, obtaining reverse replacement information corresponding to the PCR map, and obtaining the code obtained by the above decomposition. Decoding the decoded alignment value information, obtaining the alignment value information, and restoring the original data based on the reverse replacement information and the alignment value information.

The storage medium storing the compressed data of claim 11 is:

An area for storing data relating to a DC component by the DCT conversion,

An area for storing coefficients of an approximated curve obtained by arranging the AC components obtained by the DCT conversion substantially in order, and

An area for storing data relating to the following reverse replacement information Φ indicating a correspondence between the order before sorting and the order after sorting;

It has.

Φ = (Φ (Ϊ) Φ (2) Φ (3) Φ

Here, is m that satisfies (m) in the following permutation information σ.

a = (σ (1) σ (2) σ (3) σ (Μ))

Here, hi (j) indicates the number of the j-th data after sorting in the order before sorting, and M indicates the number of data.

The data transmission system of claim 12 is

It is characterized by comprising (a) a transmitting device S and (b) a receiving device.

(a) The transmitting device has the following (al) to (a6):

(al) Receives a plurality of data given in order, outputs the value indicated by each data according to a predetermined rule, outputs sorted value information, and gives the sorted order obtained by the above sorting and the given order. Means for outputting the following reverse replacement information Φ indicating the correspondence with the order

Φ = ((1) <ί> (2) 0 (3) Φ

Here, (j) is m which adds (m) ^ j in the following replacement information. σ = (σ (1) σ (2) σ (3) σ (Μ))

Here, σ① indicates the number of the j-th data after sorting in the order before sorting, and M indicates the number of data.

(a2) receiving the inverse permutation information Φ obtained by the alignment means, and outputting the following PCR map m for the inverse permutation information (>

m = [] 112,1313,, mM]

Here, the k-th element m _k is obtained by the following equation, i, V. (k) <a (k + l)

k-nk-1, V. (k)> hi (k + 1)

k = 2, 3,4, M where nk is the difference between the element with (> (a (k)) = k and the element with <K and (k + l)) = k + l in φ Is the number of elements whose value is less than k,

(a3) a PCR encoding unit that encodes the PCR map m obtained by the PCR unit and outputs an encoded PCR map;

(a4) an alignment value encoding unit that encodes the alignment value information obtained by the alignment unit and outputs encoded alignment value information;

(a5) transmitting means for transmitting the encoded PCR mapping from the PCR encoding means and the encoded alignment value information from the alignment value encoding means as compressed data,

(b) The receiving device has the following (bl) to (b5):

(b 1) receiving means for receiving the transmitted compressed data,

(b2) a PCR decoding means for receiving an encoded PCR map in the received compressed data, decoding the encoded PCR map, and obtaining a PCR map;

(b3) a reverse PCR conversion means for receiving the PCR map obtained by the PCR decryption means and obtaining reverse replacement information corresponding to the PCR mapping;

(b4) receiving an encoding alignment value information in the given compressed data, decoding the encoding alignment value information, and obtaining an alignment value information;

(b5) Restoring means for restoring the original data based on the reverse replacement information obtained by the inverse CRC conversion means and the alignment value information obtained by the alignment value decoding means.

The data transmission method of claim 13 is:

-5-Replacement paper (Rule 26) The transmitting side transmits the data after compressing the data based on the data compression method according to claim 8.

On the receiving side, data is restored based on the data decompression method of claim 10.

Claim 14 is a data storage device,

Φ = ((ΐ) (2) (3) 0 (Μ))

Here, ① is m that satisfies a (m) = j in the following replacement information.

σ = (σ (1) σ (2) σ (3) σ (Μ))

Where is the number of data after sorting: j indicates the number of data in the order before sorting, M indicates the number of data,

PCR conversion means for receiving the inverse permutation information Φ obtained by the alignment means and outputting the following PCR map m for the inverse permutation information,

Here, the k-th element m _k is obtained by the following equation.

ί Va (k) <h (k + 1)

One 1 k-i -1, V. (k)> hi (k + 1)

k = 2, 3,, where nk is the difference between the element with (> (a (k)) = k and the element with (> (a (k + l)) = k + l The number of elements whose value is less than k,

- 6 - replacement example paper (Rule ² 6) Storage means for storing the encoded mapping value information from the PCR mapping and alignment value encoding means as encoded data from the PCR encoding means,

It has.

The data storage method of claim 15 is

Receiving a plurality of data given in order, obtaining alignment value information in which the sensitivity indicated by each data is aligned according to a predetermined rule,

In order to show the correspondence between the sorted order obtained by the above sorting and the given order, order information in which the sorted order is arranged according to the given order is obtained, and the order information is obtained. Obtaining ordered mapping information structured into permutations,

Encode the alignment value information and the order mapping information to obtain and store compressed data

It is characterized by.

The data compression device of claim 17 is

The compression model selection means and the compression model selection means for receiving a plurality of data given in order and selecting whether to perform permutation coding or non-permutation coding based on the value indicated by the plurality of data. Then, when the permutation coding is selected, the compression by the permutation coding means (a) described below is executed, and when the non-coding method is selected by the compression model selection means, the following (b) Compression execution means for executing compression by the non-permutation encoding means of

(a) To obtain sorting value information in which values indicated by given data are sorted according to a predetermined rule, and to indicate a correspondence between the sorted order obtained by the sorting and the given order. , Obtains order information in which the order after sorting is arranged according to the given order, obtains order mapping information in which the order information is structured in a permutation, encodes and compresses the sort value information and the order mapping information, respectively. Locator to obtain data,

(b) When non-encoding is selected by the compression model selection means, non-permutation is performed in which the given data is encoded to obtain compressed data without performing the above-described permutation encoding. Encoding means,

It has. The data compression apparatus according to claim 18 is characterized by further comprising orthogonal transformation means for orthogonally transforming the given original data and providing the orthogonal data to the compression model selecting means. A data compression apparatus according to claim 19 is characterized in that the orthogonal transformation means receives the original data, performs DCT transformation, obtains DCT transformation data obtained by zigzag scanning, and provides the data to the compression model selection means.

A data compression apparatus according to claim 20, wherein the DCT transform data received from the orthogonal transform means is quantized according to a predetermined quantization roughness, and the DCT quantized by the quantizing means. Effective data extraction means for extracting, from the converted data, data having the lowest frequency component to data having the lowest frequency component exceeding a predetermined value as effective coefficient data, and providing the effective coefficient data to the compression model selecting means. I have.

The data compression apparatus according to claim 21 is characterized in that the compression model selecting means includes whether or not the maximum value of the AC component of the DCT conversion data is smaller than a predetermined value as a criterion for selecting the compression model. I have.

A data compression apparatus according to claim 22 is characterized in that the compression model selecting means includes whether or not the number of effective coefficients is smaller than a predetermined value in a criterion for selecting a compression model.

Claim 23 is a data compression device,

The encoding means includes:

Receiving a plurality of data given in order, outputting the value indicated by each data in accordance with a predetermined rule, outputs sorted value information, and comparing the ordered order obtained by the above sorting with the given order. An alignment method that outputs the following replacement information σ indicating the correspondence,

σ = (σ (1) σ (2) σ (3) σ (Μ))

Here, σ (j) indicates the number of the j-th data after sorting in the order before sorting, and M indicates the number of data.

PCR conversion means for receiving the permutation information σ obtained by the alignment means and outputting the following Ρ C R map m for the permutation information σ:

Here, the k-th element m _k is obtained by the following equation. n _k , V (k) ku (k + 1)

k-iik-1, V. (k)> a (k + l)

k = 2, 3, 4, M where _mk is the value of σ between the element with a (k) = k and the element with a (k + l) = k + l the number of elements less than k,

A PCR encoding unit that encodes the PCR map m obtained by the PCR unit and outputs an encoded PCR map;

Output means for outputting the encoded PCR mapping from the PCR encoding means and the encoded alignment value information from the alignment value encoding means as compressed data;

It is characterized by having.

The data compression device according to claim 24,

The replacement encoding means,

Receiving a plurality of data given in order, outputting the value indicated by each data in accordance with a predetermined rule, outputs sorted value information, and comparing the ordered order obtained by the above sorting with the given order. Alignment means for outputting the following reverse replacement information 0 indicating the correspondence,

Φ = (Φ (ΐ) Φ (2) (3) Wan

Here, φφ is m that satisfies a (m) = j in the following permutation information.

σ = (σ (1) σ (2) σ (3) σ (Μ))

In response to the reverse permutation information 0 obtained by the alignment means, the reverse permutation information (PCR conversion means for outputting the following PCR map m,

m2 [ni2, m3,, mMj ^T

Here, the k-th element m _k is obtained by the following equation.

-9-Replacement paper (Rule ²⁶⁾ = U. (k) <h (k + 1)

^-I kn _k -l, V. (k)> HI (k + 1)

k = 2, 3, 4 M where, ilk is the difference between the element with () (a (k)) = k and the element with (/) (a (k + l)) = k + l PCR encoding means for encoding the PCR map m obtained by the PCR encoding means, the number of which is smaller than k, and outputting an encoded PCR mapping,

Output means for outputting the encoded PCR mapping from the PCR encoding means and the encoded alignment value information from the aligned and direct encoding means as compressed data;

It is characterized by having.

The data compression device according to claim 25, wherein the alignment value encoding means performs encoding by expressing a value obtained by multiplying the alignment value by an expansion coefficient as a function in encoding the alignment value. Features.

A data compression apparatus according to claim 26 is characterized in that the non-permutation encoding means encodes data based on the JPEG standard.

The data compression method of claim 27 is

Receives a plurality of data given in order and determines whether to perform permutation coding or non-permutation coding based on the value indicated by the plurality of data,

When the permutation coding is selected, the compression by the permutation coding shown in (a) below is executed, and when the non-permutation coding is selected by the compression model selecting means, the non-permutation coding shown in (b) below is performed. Is performed to obtain compressed data.

(a) If the permutation coding is selected, the sorted value information obtained by sorting the values indicated by the given data according to a predetermined rule is obtained, and the sorted order obtained by the above sorting is obtained. In order to show the correspondence between the order and the given order, order information in which the sorted order is arranged according to the given order is obtained, and the order place mapping information in which the order information is structured in the order is obtained. To encode the alignment value information and the order mapping information, respectively, to obtain compressed data.

-10-Replacement form (Rule ²⁶ ) (b) When non-permutation coding is selected, the given data is coded to obtain compressed data without performing the above-described constraining.

The data decompression device of claim 28 is

Compression model determining means for determining, based on the compression model data in the given compression data, the force of the compression processing by the permutation coding and the compression model by the non-permutation coding;

If it is determined that the encoding is performed, decompression is performed by the permutation code decompression means shown in (a) below. It is characterized by performing decompression.

(a) If it is determined that the rice is F ^^-encoded, an encoded PCR map is obtained from the given compressed data, and the encoded PCR map is decrypted to obtain a PCR map. Obtaining reverse permutation information corresponding to the mapping, obtaining encoded alignment value information from the given compressed data, decoding the encoded rearrangement information, obtaining alignment value information, and obtaining the reverse permutation information. Permutation decoding means for restoring data based on and the alignment value information,

(b) Non-permutation decoding means for decoding the given compressed data and restoring the data without performing the above-mentioned permutation decoding when determining that the encoding is non-permutation encoding.

The data decompression method of claim 29 is as follows:

Based on the compression model data in the given compressed data, it is determined whether or not the compression processing by the permutation coding has been performed and whether the compression processing by the non-permutation coding has been performed.

If it is determined to be permutation encoding, decompression by decompression by permutation code of (a) below is executed, and if it is determined to be non-permutation encoding, decompression by decompression by non-permutation code of (b) below It is characterized by performing.

(a) If it is determined that the encoding is permutation encoding, an encoded PCR map is obtained from the given compressed data, the encoded PCR map is decoded, and a PCR map is obtained. To obtain the encoded alignment value information from the given compressed data, and obtain the encoded alignment value information. To obtain alignment value information, and restore data based on the reverse replacement information and the alignment value information;

(b) If it is determined that the data is a non-permuted code, the given compressed data is decoded to restore the data without performing the above-described replacement decoding.

The storage medium of claim 30 stores compressed data having the following structure: a compressed model storage area for storing a pressure box model;

When the compression model stored in the compression box model storage area is the permutation coding, it has the following areas (a) to (c),

(a) an area for storing data relating to a DC component by the DCT conversion,

(b) an area for storing a coefficient of an approximate curve obtained by arranging AC components due to the DCT transformation in a substantially large order,

(c) Bii¾ An area for storing data indicating the correspondence between the order before sorting and the order after sorting If the compression model stored in the compression model storage area is non-permutation coding, the following (~ ( e) the area

(d) a territory that stores data relating to the DC component obtained by the DCT conversion,

(e) An area for storing data on AC components obtained by DCT conversion without rearranging the order.

The data transmission system of claim 31 is characterized by comprising: (a) a transmitting device and (b) a receiving device.

(a) The transmitting device has the following (al) to (a3):

(al) a compression model selecting means for receiving a plurality of data given in order, and selecting whether to perform permutation coding or non-permutation coding based on the value indicated by the plurality of data,

(a2) When the permutation encoding is selected by the compression model selecting means, the compression by the following permutation encoding means (a21) is executed, and when the non-permutation encoding is selected by the compression model selecting means, Is compression executing means for executing compression by the non-permutation encoding means of (a22) below,

(a21) Obtain sorting value information in which the values indicated by the given data are sorted according to a predetermined rule, and compare the sorted order obtained by the sorting with the given order. In order to show the correspondence, in accordance with the given order, order information in which the order after sorting is arranged is obtained, and the order information is obtained by arranging the order information in a permutation. A permutation encoding means for respectively encoding and obtaining compressed data;

(a22) When non-replacement encoding is selected by the compression model selection means, non-permutation is performed in which the given data is encoded to obtain compressed data without performing the above-described permutation encoding. Encoding means,

(a3) transmitting means for transmitting compressed data from the compression executing means,

(b) The receiving device has the following (bl) to (b3):

(bl) receiving means for receiving the transmitted compressed data,

(b2) a compression model determining means for determining, based on the compression model data in the received compressed data, whether the compression processing by the transcoding is performed and whether the compression processing by the decoding is performed;

(b3) If it is determined that the encoding is permutation encoding, the decompression by the permutation code decompression means described in (b31) below is executed. A data decompression device that performs decompression by code decompression means;

0> 31) If it is determined that the encoding is permutation encoding, an encoded PCR map is obtained from the given compressed data, the encoded PCR map is decoded, and a PCR map is obtained. Obtains the encoded alignment value information from the given compressed data, decodes the encoded alignment value information, obtains the alignment value information, and obtains the inverse replacement information and the alignment value information. A permutation decoding means for restoring data based on

Φ32) Non-permutation decoding means for decoding the given compressed data and restoring the data without performing the above-described permutation decoding when it is determined to be non-permutation encoding.

Claim 32 is a data storage device,

The compression model selection means and the compression model selection means for receiving a plurality of data given in order and selecting whether to perform permutation coding or non-permutation coding based on the value indicated by the plurality of data. If the replacement encoding is selected, the replacement of (a) below A compression execution means for executing the compression by the encoding means and executing the compression by the non-permutation encoding means of the following (b) when the non-permutation encoding is selected by the compression model selecting means;

(a) Obtain sorting value information in which the values indicated by the given data are sorted in accordance with a predetermined rule, and provide a value to indicate the correspondence between the sorted order obtained by the above sorting and the given order. In addition to obtaining order information in which the order after sorting is arranged according to the specified order, obtaining order mapping information in which the order information is structured in a permutation, encoding the sort value information and the order mapping information, and compressing the compressed data. Permutation encoding means,

(b) If non-permutation encoding is selected by the compression model selection means, non-permutation encoding is performed, in which the given data is encoded to obtain compressed data without performing the above-described permutation encoding. Means,

Storage means for storing the compressed data;

It has.

Claim 33 The data storage method of claim 3 is

When the permutation coding is selected, the compression by the permutation coding shown in (a) below is executed, and when the non-permutation coding is selected by the compression model selecting means, the non-replacement coding shown in (b) below is performed. It is characterized in that the compressed data is obtained by executing a compression box by encoding, and the obtained compressed data is stored.

(a) When the permutation coding is selected, the value indicated by the given data is sorted according to a predetermined rule to obtain sorting value information, and the order after the sorting obtained by the above sorting is given. In order to show the correspondence with the given order, in accordance with the given order, we obtain order information in which the sorted order is arranged, and obtain order place mapping information in which the order information is structured in order. Encoding the alignment value information and the order mapping information to obtain compressed data,

(b) When non-replacement encoding is selected, given data is encoded without performing the above replacement encoding to obtain compressed data. In the present invention, "the sorting means J means to sort the value indicated by each data in accordance with some rule such as ascending order, descending order and the like. In the embodiment, step S40 in FIG. 7 corresponds to this.

"Multiple data given in an ordered manner" is a concept that refers to the case where data ordered in some form is given. This includes cases in which data ordered according to is given.

"Aligned value encoding means" means means for encoding the values of sorted data, encoding by function such as curve approximation, encoding focusing on repetition regularity, run-length encoding And the like. In the embodiment, steps S40 to S48 in FIG. 7 correspond to this.

“Output means” refers to means for providing compressed image information to the own device or another device. In the embodiment, step S22 in FIG. 6 corresponds to this.

The “permutation encoding means” refers to a means for performing encoding processing by separating into sorting value information obtained by sorting and sorting information in sorting. In the embodiment, Step S 18 in FIG. 6 (detailed in FIG. 7) corresponds to this.

「The“ program ”in the storage medium j that stores the program refers to not only the case where the program can be directly executed on a computer, but also the case where the program can be executed in cooperation with another program or OS, or This is a concept that includes cases that can be executed later.

The data compression apparatus according to claim 1, 2 or 3, receives a plurality of data given in an order, sorts the value indicated by each data in accordance with a predetermined rule, and sort value information. It obtains a PCR map of inverse permutation information indicating the correspondence between the obtained order after the alignment and the given order, and outputs the encoded information as compressed data. Since the PCR mapping for the reverse replacement information is encoded, the compression ratio can be further improved.

The data compression device according to claim 4 is effective in quantizing DCT transform data from the lowest frequency component to the highest frequency component exceeding a predetermined value. They are extracted as coefficients and given to the sorting means as sorting target data. Therefore, the effective coefficient can be extracted quickly.

The data compression device according to claim 6 is characterized in that the alignment value encoding means performs encoding by expressing the alignment value as a function. Therefore, the compression ratio can be improved.

The data compression device according to claim ⁷ is characterized in that the alignment value encoding means performs encoding by expressing a value obtained by multiplying the alignment value by an expansion coefficient as a function. Therefore, the quantization error can be reduced.

The data compression method according to claim 8 receives the plurality of data given in order, obtains sorted value information in which values indicated by the respective data are sorted in accordance with a predetermined rule, and obtains sorted data obtained by the sorting. In order to show the correspondence between the order and the given order, in addition to obtaining the order information in which the sorted order is arranged in accordance with the given order, obtaining the order mapping information in which the order information is structured into a permutation, It is characterized in that the value information and the order mapping information are encoded respectively to obtain compressed data. That is, since the sorted alignment value information is encoded, the compression ratio can be improved. A data storage device according to claim 14 receives a plurality of data given in an order, sorts the value indicated by each data according to a predetermined rule, and sort value information obtained by the sorting. A PCR mapping of the inverse replacement information Φ indicating the correspondence between the later order and the given order is obtained, and these encoded information are output as compressed data. Since the PCR mapping for the reverse replacement information is encoded, the compression ratio can be further improved and the storage efficiency can be improved.

A data storage method according to claim 15, further comprising: receiving a plurality of data given in an order, obtaining sorting value information in which values indicated by the respective data are sorted according to a predetermined rule, and obtaining the sorting value obtained by the sorting. In order to show the correspondence between the later order and the given order, in order to obtain the order information in which the sorted order is arranged according to the given order, and obtain the order mapping information in which the order information is structured in a row, It is characterized in that the alignment value information and the order mapping information are respectively encoded, and compressed data is obtained and stored. Since the aligned alignment value information is encoded, a high compression rate can be obtained. The data compression device according to claim 17, wherein the values indicated by the plurality of data given in order Based on this, whether to perform permutation coding or non-permutation coding is selected. Therefore, given data can be more appropriately and efficiently encoded, and the compression ratio can be improved.

A data compression apparatus according to claim 21 is characterized in that whether or not the maximum value of the AC component of the DCT conversion data is smaller than a predetermined value is included in the criterion for selecting a compression model. Therefore, an appropriate coding model can be selected according to the maximum value of the AC component.

The data compression device according to claim 22 is characterized in that whether or not the number of effective coefficients is smaller than a predetermined value is included in the criterion for selecting a compression model. Therefore, an appropriate coding model can be selected according to the number of effective coefficients. BRIEF DESCRIPTION OF THE FIGURES

FIG. 1 is a diagram showing an overall configuration of a compression device according to an embodiment of the present invention. FIG. 2 is a diagram showing an entire configuration of a defrosting apparatus according to one embodiment of the present invention. FIG. 3 is a diagram showing a hardware configuration in a case where the devices g of FIGS. 1 and 2 are realized using CPU.

FIG. 4 is a diagram illustrating an example of a block of image data.

FIG. 5 is a diagram showing a flowchart of the compression processing program.

FIG. 6 is a diagram showing a flowchart of the compression processing program.

FIG. 7 is a diagram showing a flowchart of the compression processing program.

FIG. 8 is a diagram showing a zigzag scan.

FIG. 9 is a diagram showing original block image data and DCT coefficients.

FIG. 10 is a diagram showing a quantum roughness table.

FIG. 11 is a diagram for illustrating processing by permutation encoding.

FIG. 12 is a diagram for illustrating permutation, reverse permutation, and PCR mapping.

FIG. 13 is a diagram showing DCT coefficients and image data obtained by decompression.

FIG. 14 is a diagram showing a data structure after compression.

FIG. 15 is a detailed diagram showing the data structure after compression.

FIG. 16 is a detailed diagram showing the data structure after compression. FIG. 17 is a diagram showing a table of the compression model.

FIG. 18 is a diagram showing an outline of the conversion to F-number.

FIG. 19 is a diagram showing a flowchart of the decompression processing program.

FIG. 20 is a diagram illustrating a restoration process from the PCR mapping to the reverse permutation.

FIG. 21 is a graph comparing J PEG with the compression according to the present invention.

Figure 22 shows the image Lena used for comparison and its luminance histogram. FIG. 23 shows an image Lena subjected to the compression and decompression processing of the present invention and its luminance histogram.

Figure 24 shows an image Lena and its luminance histogram after JPEG compression and decompression processing.

FIG. 25 is a diagram showing an overall configuration of a compression device according to another embodiment.

FIG. 26 is a diagram showing an entire configuration of a defrosting device according to another embodiment. Embodiment of the Invention

FIG. 18 shows a main part of the compression processing by the permutation encoding according to the present invention. Figure 188 shows the target data. Preferably, the data to be compressed has a tendency to be arranged from a large value to a small value if possible. Therefore, when the original data does not have such a tendency, it is preferable to perform the orthogonal transformation or the like and use the data having such a tendency as the target data.

First, the target data is sorted within the range of roughness D, allowing the size to be reversed. As a result, the data is sorted as shown in Fig. 18B, and sorted data (sorted value information) is obtained. Curve approximation is performed on data that exceeds a predetermined threshold value among the target data sorted and rearranged. As a result, even a large amount of data can be replaced with a very small number of curve approximation coefficients.

On the other hand, as the permutation information for the rearrangement, the permutation information σ (obtains the force at the position of the original data. Further, the inversion 换 information (at what position of the rearranged data the original data The PCR map of the reverse permutation information is obtained, and this is subjected to Huffman coding.

Convert the curve approximation coefficients obtained above and the Huffman-encoded PCR mapping into compressed data And

FIG. 1 shows an overall configuration of a data compression device according to an embodiment of the present invention. The image information given to the blocking means 1 is divided into blocks of 8 pixels vertically and 8 pixels horizontally. The blocked image information is provided to the two-dimensional DCT means 2 and converted into two-dimensional DCT conversion data. The two-dimensional DCT data thus obtained is zigzag scanned by the zigzag scanning means 4 from the side of the lower frequency component. The quantization means 7 quantizes this with a predetermined quantization roughness, and gives it to the effective coefficient extraction means 5. Effective coefficient extraction means 5 extracts effective coefficients to be subjected to compression processing from the quantized two-dimensional DCT data.

The compression model selection means 9 selects whether to perform permutation coding or non-permutation coding based on the number and value of effective coefficients. When non-permutation coding is selected, the effective coefficient is coded and compressed by the J PEG means 21 in the same manner as J PEG. The output means 20 outputs the compression model ゃ the coded effective coefficient and the like as compressed image information.

When the permutation encoding is selected, the compression processing is performed by the permutation encoding means 8. The replacement encoding means 8 includes an alignment means 6, an alignment value encoding means 10, a PCR means 16 and a Huffman encoding means 18. The sorting means 6 sorts the effective coefficients roughly in descending order of amplitude according to a predetermined roughness. As a result, the sorted alignment value information and the information on the permutation of the data order accompanying the sorting are obtained. As information for changing the order of the data accompanying the sorting, replacement information, reverse replacement information, and the like are used. The alignment value information is encoded by the alignment value encoding means 10 and output as encoded alignment value information. On the other hand, the reverse replacement information obtained by the alignment means 6 is given to the PCR conversion means 16 and the PCR map is calculated. This PCR map is Huffman-encoded by Huffman encoding means 18 and output as an encoded PCR map. The output means 20 outputs a compression model, the above-described encoded alignment value information, an encoded PCR map, and the like as compressed image information.

FIG. ² shows an overall configuration of a data decompression device according to an embodiment of the present invention. The given compressed image information is determined by the compression model determination means 22 as to whether the compression processing has been performed by replacement coding or non-replacement coding. If it is determined that the compression is performed by de-encoding, the effective coefficient is decoded by the inverse JPEG means 23 and given to the inverse zigzag scanning means 36.

When it is determined that the compression is performed by the permutation encoding, the effective coefficient is decoded by the non-decoding means 29. The non-permutation decoding means 29 includes a Huffman decoding means 28, an inverse PCR means 30, an alignment value decoding means 40, and an alignment cancellation means 34. The alignment value decoding means 40 receives the encoded alignment value information in the compressed image information and decodes it into alignment value information. On the other hand, the Huffman decoding means 28 receives the coded PCR mapping in the compressed plane image information and decodes it into a PCR mapping. The inverse PCR conversion means 30 converts this PCR map into inverse replacement information and outputs it.

The alignment value information and the reverse replacement information obtained as described above are provided to the alignment canceling means 3. The sort canceling means 34 receives the sort value information and the reverse replacement information, and obtains quantized effective coefficients in the order before sorting. The inverse quantization means 35 inversely quantizes this to obtain an effective coefficient.

The inverse zigzag scanning means 36 receives the effective coefficient from the unsubstituted decoding means 29 or the inverse JPEG means 23 and converts it into two-dimensional DCT data. Further, the inverse two-dimensional 13 (the means 38 obtains the original blocked image information based on the two-dimensional DCT data. The inverse blocker 39 obtains the original image based on the blocked image information. get information.

FIG. 3 shows a hardware configuration when the data compression apparatus in FIG. 1 and the data decompression apparatus in FIG. 2 are realized using a CPU. A hard disk 52, a display 54, a communication circuit 56, a memory 58, a floppy disk drive (FDD) 60, a keyboard and a mouse 62 are connected to the CPU 50. The hard disk 54 stores a program 68 for compression and a program 70 for decompression. The original data to be compressed is stored as an original data file 72, and the compressed compressed data is stored as a compressed data file 74. The compression program 68 and the decompression program 70 are installed on the hard disk 52 from the floppy disk 64 via the FDD 60. Note that the program may be installed from another storage medium such as a CD-ROM. The image data to be compressed is supplied from the floppy disk 66, 2 to be the original data file.

In the data compression according to this embodiment, one image data is stored in a block B as shown in FIG. .. The blocks are divided into blocks B to B, and compression processing is performed with these blocks as one unit. Here, one block is 8 pixels vertically and 8 pixels horizontally. In addition, a row of blocks (for example, B., .. ~ B ..) is grouped into slices. In the figure, slices 0 to P are grouped.

FIG. 5 is a flow chart showing the Kurihara program 68. First, in step S1, a pressure box parameter is set and stored in the Head section of the compressed data file 74 as shown in FIG. Here, the compression parameters are the number of lines NOL of one image, the number of pixels per line PPL, the number of colors NCH (1 for black and white, 3 for color), the number of bits per pixel BPP, 1 The number of rows in the block is SOB, and the number of the roughness table is SFC. These parameters may be set each time, but are preferably set in advance as default values.

Next, the CPU 50 executes block B. .. Two-dimensional DCT processing is performed on the 64 pixels (8 vertical pixels and 8 horizontal pixels) of the target (step S2). An example of the resulting DCT coefficient is shown in Figure 9B. FIG. 9A shows the original image data.

Next, the DC coefficient among the obtained DCT coefficients is quantized according to the roughness of the compression parameter by the roughness table (step S3). Figure 10 shows an example of a roughness table. A quantization roughness is set for each of the DCT coefficients. Since "16" is set for the DC coefficient, quantization is performed using this. In the example of FIG. 9, since the DC coefficient is “305”, the quantized DC coefficient is “19”. Next, the difference between the quantized DC coefficient of the immediately preceding block and the quantized DC coefficient obtained this time is calculated (step S4). Assuming that the quantized DC coefficient of the immediately preceding block is "64", this difference value is "1-45". In step S5, the calculated difference value of the quantized DC coefficient is stored in a buffer (step S5).

Next, the DCT coefficients excluding the DC coefficient (that is, A. coefficient) are zigzag-scanned as shown in FIG. 8 and arranged in ascending order of frequency (step S55). Figure 9 For example, the following data is obtained.

(1) (2) (3) (4) (5) ..... (63)

AC coefficient -52-118 3 55 -97 ..... 1

Among the AC coefficients arranged in this way, those to be processed are extracted as effective coefficients (steps S6 to S9). This is because the AC coefficient of the higher frequency component tends to have a smaller value. In addition, the AC coefficients are set in the order of zigzag scan from low frequency components to high frequency components.

First, the AC coefficient having the highest frequency component among the arranged AC coefficients (that is, the last AC coefficient (63)) is set as the target AC coefficient (step S6). In the example of FIG. 9, “1” at the lower right is the attention coefficient. Next, the target AC coefficient (63) is quantized by a predetermined quantization roughness based on the following equation to calculate a quantized AC coefficient (step S65).

Quantized AC coefficient = noteworthy AC coefficient / quantity roughness Note that decimals are rounded off. Further, as shown in FIG. 10, the quantization roughness is set for each AC coefficient in ascending order of frequency. Also, the higher the frequency, the lower the degree to which the human eye perceives the spatial frequency. Therefore, the higher the frequency, the greater the quantization roughness.

The CPU 50 determines whether the quantized AC coefficient calculated in step S65 is “0” or not. Here, the AC coefficient (63) is "1", and the quantization roughness for it is "79", so the quantized AC coefficient is "0". Therefore, the process proceeds to step S8. .

In step S8, the AC coefficient immediately before (low in frequency) is set as the noted AC coefficient. Here, the AC coefficient (62) is the target coefficient. Similarly, it is determined whether the quantized AC coefficient is "0". If it is "0", the process repeats with the AC coefficient one before this as the noted AC coefficient.

In the repetition of the above process, when the quantized AC component is not "0", the flow advances to stearyl-up S ^9. In step S9, the effective coefficients from the first AC coefficient (1) to the noted AC coefficient are set. The number of effective coefficients is defined as the effective coefficient length E length. Further, the remaining AC coefficients are set as invalid coefficients (step S10). Next, the effective coefficients are quantized based on the respective quantization roughnesses (see Fig. 10) to obtain quantized AC coefficients (quantized effective coefficients) (step SI1). An example of the quantized AC coefficient calculated in this way is shown in “Signed Quantized Coefficient” in FIG. In this example, the AC effective coefficient length Elength is "25".

Next, the CPU 50 finds the quantized AC coefficient having the largest absolute value among the obtained quantized AC coefficients, and sets it as the maximum quantized AC coefficient XS1 (step S12). In the example of FIG. 11, XS 1 is “1 12”. Further, the number of non-zero quantized AC coefficients is calculated, and this is set as the effective effective coefficient length RE length (step S13). In the example of FIG. 11, RE length is "14". This effective coefficient length R E lengt is Huffman coded and stored in a buffer.

Next, for the quantized AC coefficients that are not "0", they are separated into absolute values and positive and negative signs (step S15). Further, a bit string indicating a positive / negative sign (a bit string with positive being 0 and negative being 1) is stored in the buffer (step S16).

Subsequently, in step S17, it is determined whether or not the calculated XS1 is "7" or less and the RE length is "9" or less. If this condition is not satisfied, compression processing by permutation encoding is performed (step S18). If this condition is satisfied, compression processing by JPEG processing, which is non-permutation encoding, is performed (step S19). This is because when the maximum quantized AC coefficient XS 1 is small, the power and the effective effective coefficient length R E length are small, the compression ratio using the approximated curve is not so high.

First, processing (JPEG processing) when XS 1 is equal to or less than “7” and RE length is equal to or less than “9” will be described. Figure 7 shows a flowchart of this JPEG processing. In this case, first, in order to indicate that JPEG processing has been performed on the block, the value of the compression model model is stored as "0" in the buffer (step S30). Next, the quantized AC coefficient (effective coefficient) is encoded using a codebook for run-length Huffman encoding in the same manner as JPEG, and is stored in a buffer (step S31). Since the maximum run length of the zero coefficient does not exceed "9" and the maximum value of the non-zero coefficient does not exceed "7", it is necessary to generate a code with fewer bits than JPEG. Can be. Next, processing (replacement encoding) when XS 1 is not equal to or less than “7” or when RE length is not equal to or less than “9” will be described. Figure 7 shows a flowchart of this substitution coding.

First, the absolute values of the quantized AC coefficients (effective coefficients) are arranged in descending order (step S40). In other words, this means that the AC coefficients are sorted in ascending order of absolute value, using the quantization roughness as the sort roughness. The sorted quantized AC coefficients are called alignment values (see “ΓAlignment values” in Fig. 11). When sorting, the CPU 50 stores the following replacement data indicating the relationship between the original order and the sorting order (see FIG. 11).

HI = (σ (1) a (2〉 · ..σ (Μ))

Here, σ θ) indicates the number of the j-th data after sorting in the order before sorting. M is the effective coefficient length Elength.

Next, an approximate curve is obtained for the effective coefficients for which the quantized AC coefficient is not 0 (steps S41 to S42). First, in step S41, a logarithmic expanded value Y (p) is obtained for a quantized AC coefficient that is not 0.

Υ (ρ) = 36 * 1η (| Ζ (σ (ρ)) / θ (σ (ρ)) Ι)

ρ = 1,, RElength where Ζ (ί) is the AC coefficient and D (i) is the corresponding quantization roughness (see Fig. 10).

Z (R ZD (i) usually decreases exponentially. Therefore, by approximating the logarithm, the approximate curve can be approximated to a straight line. It is multiplied by a factor (36) in order to reduce the error when approximating as a function by increasing the value.The logarithm obtained for the alignment values shown in Figure 11 The generalized expansion value Y (p) is shown below.

Y = (898963615236302518 -9 -9 -21 -17 -5) An approximate curve is calculated by the least squares method for the logarithmic expanded value calculated in this way (step S42). In this embodiment, it provides the following three models, seeking minimum error Q ² become coefficients j3 using least-squares method for each model. Model 1: QM) = EJyW (l) f (l, s)] ² f (l, s) = ln (s) Model 2: Qi (M) = E ^M [y (s)-^ (2) f (2 _(S ) P f (2, s) = s-1 Model 3: E = E ^M [y (s) -¾t) f (t, s)] ²

Select the model error Q ² is equal to or less than a predetermined threshold value, to obtain the coefficients of the approximate curve. The models are examined in the order of 1, 2, and 3. When an error with an error equal to or less than the threshold is obtained, the relevant model is selected without examining the other models. In this way, the processing speed has been improved. If the error does not fall below the threshold in any of the models, the model with the smallest error is selected. Alternatively, the operation may be performed again after changing the quantization roughness. As described above, the model number and the coefficient in the model are calculated. This allows a lot of data to be represented by a model number and a few coefficients, resulting in a high compression ratio.

When the data in Fig. 11 is approximated by a curve, the model = 2 and the coefficient Qa2 = 16 is obtained. In the case of model 1, the coefficient Q al is obtained, and in the case of model = 3, the coefficient Q al and the coefficient Q a2 are obtained. The obtained model number and curve approximation coefficient are stored in the buffer (step S43). In addition, it calculates the approximation error between the quantized AC coefficient approximated by the curve approximation coefficient and the true quantized AC coefficient (see “Approximation Error” in Figure 11). Store the calculated approximation error in the buffer

(Step S44). After that, store the maximum value of AC coefficient Xsl in the buffer

(Step S45).

After the curve approximation is obtained as described above, the following inverse replacement data is calculated for the replacement data σ obtained in step S40 (step S46). σ = (σ (1) σ (2) .... σ (M))

(> = ((> (1) φ (2) .... φ (Μ))

Here, (j) is m that satisfies σ (m) = j in the replacement data. Figure 11 shows the inverted data. Here, (j) indicates the order of the j-th data before sorting in the order after sorting. Since the reverse replacement data can be coded more efficiently than the replacement data, this is an object of coding in this embodiment.

In this embodiment, the reverse data is calculated after the replacement data σ is calculated. However, the reverse data may be calculated directly at the time of sorting.

Next, the CPU 50 calculates a PCR (Permutation Cyclic Representation) mapping m of the reverse replacement data (step S47).

V. (k) <h (k + 1)

.k-iik-1, Va (k)> a (k + l)

k = 2, 3, 4 M wherein, m _k is (> in, () (a (k)) is a = k element and <> (a (k + l )) = a k + l elements Is the number of elements whose value is less than k (see Figure 12), where [] ^τ indicates the transpose of the matrix.

For inverse permutation (this in the case of FIG. 1 2, the procedure of calculating the element m ₂ of PCR mapping m (that is, if the k = 2) are as follows. First, from among the inverse permutation, "2 In this case, it indicates the position of σ (2) of the permutation σ from "5" and the position of the inverse permutation ("2") from the beginning. In other words, referring to the permutation σ, The position (5th) of the reverse permutation (the second) can be specified. Next, the position (1st) of the "3" in the reverse permutation is specified in the same manner. Since "3" precedes "" (since σ (1> σ (k + 1)), m ₂ is obtained as kn ₂ -l. between "2" and "3", not smaller than "2" element is only one of "1" Therefore, kn ₂ -. 1 is a 2-1-1, m ₂ is It becomes "0".

m ₃ can be calculated similarly. However, reverse substitution (M is "4" rather than "3")

-26-Replacement form (Rule 26) Is behind, so m ₃ is found as na. Thus, ma is "1", the number of elements smaller than "3" in the question "3" and "4" (see Figure 12). The PCR map m calculated as described above is shown below.

m = [0110 -211 -20200254200000001] ^τ Next, run-length Huffman coding is performed on the calculated PCR mapping and stored in a buffer (step S48). In general, PCR mapping has the property that successive zero elements and non-zero elements alternate. Therefore, by converting to a PCR mapping, efficient coding can be performed using run-length 'Huffman coding. In this way, the information on the sorting of the effective coefficients can be encoded.

When the data compressed by the permutation encoding or the JPEG processing is obtained as described above, it is determined in step S20 in FIG. 6 whether or not the processing has been completed for all the blocks (step S20). . If the processing has not been completed, the next block is set as a target block (step S21), and steps S2 and subsequent steps are repeatedly executed to store compressed data in the buffer.

When the processing is completed for all the blocks, the contents of the buffer are recorded as a compressed data file on the hard disk 52 or a floppy disk or the like (step S22). When transmitting, the contents of the buffer are transmitted via the communication circuit 56.

Here, the structure of the compressed data is shown in FIG. Following the head section that stores the compression parameters for one image data, the data of each slice section (see FIG. 4) that constitutes one image data is stored. In each slice, data of a block constituting the slice is stored.

Fig. 15 shows the details of the head, and Fig. 16 shows the details of the block. A block unit, model number _mo del is stored, thereby the proc is summer so it can be determined whether J PEG coding substituted coding. Compressed data is stored as a different data structure when model = 0 and when model = 1, 2, and 3.

First, the difference value diHQDC of the quantized DC coefficient calculated in step S5 is described. Has been recorded. Next, the effective effective coefficient length RElengtli is recorded, the sign of the effective coefficient is recorded, and the model model number is recorded.

In the case of moedl = l, 2, or 3, the approximate curve coefficient calculated in step S42 is recorded, the maximum value xsl of the effective coefficient is recorded, and the approximate error calculated in step S44 is recorded. The CFE is recorded, the Huffman-encoded PCR mapping calculated in step S48 is recorded, and the EOB indicating the end of the block is recorded.

In the case of model = 0, data obtained by encoding the quantized effective coefficients by the codebook according to the JPEG system is stored, and an EOB indicating the end of the block is recorded.

By replacing such a compressed data file with the original data file of the floppy disk 66, the substantial storage capacity of the floppy disk 66 can be increased. Also, when storing data on a hard disk or the like, storing the compressed file increases the actual storage capacity. Furthermore, when data is transmitted via the communication line 56, the transmission time can be shortened by transmitting the data after compression. If compression and decompression are performed by different devices, they must have the same model table (Fig. 17) and quantization roughness table (Fig. 10).

FIG. 19 shows a flowchart of the decompression program 70 for restoring the compressed data. First, in step S50, the compression parameters (FIG. 15) of the head are read, and the parameters for the single image are restored. Next, the data of the first block is read (step S51). Next, based on the difference diffQDC between the quantized DC coefficient of the previous block and the quantized DC coefficient, the quantized DC coefficient of the current block is calculated. Further, based on the quantized DC coefficient, the DC coefficient is restored using the quantization roughness of the DC coefficient stored in the quantization roughness D table (see FIG. 10) (step S52).

Next, the effective sign length RElengtlu RElength sign signs and mod el are obtained (steps 53, 54, 55).

In step S56, it is determined whether or not the acquired model is "0" (that is, PC orchids 60

It is determined whether the compression is performed by substitution coding or compression by JPEG (step S56). If model = 0, it is determined that the encoding is performed by JPEG, and the quantified effective coefficient is decoded using an associated codebook to obtain a quantized effective coefficient (step S57). .

On the other hand, if it is determined in step S56 that model = 1, 2, or 3, the process from step S57 is performed. First, RElength quantized AC coefficients are calculated based on the model number model, the approximate curve coefficient, and the near error (step S58). Next, the Huffman-coded PCR mapping is decoded to obtain the original PCR mapping (step S59). The reverse permutation information </> is obtained from the obtained PCR mapping, and the permutation information σ is further obtained from the reverse permutation information (step S60).

Here, a specific process at the time of the reverse substitution information (this conversion will be described from the PCR map. Here, the PCR map m shown in FIG. 12 will be described with reference to FIG. 20. First, (1 Next, cyclically shift to the left by the value of m ₂ (that is, 0) (shift right if the value is negative), which gives (1 2) On the other hand, add "3" at the end to obtain (1 23). Then, cyclically shift to the left by the value of m ₃ (that is, 1). On the other hand, "4" is added at the end to obtain (23 1 4). Such a process can be repeated to obtain reverse substitution information. Conversion to the inverse permutation information φ can be easily obtained by the shift operation, so that decompression can be performed quickly.

Next, according to the replacement information obtained in this way, the sort of the quantized AC coefficient is canceled (step S61). Here, Elength quantized AC coefficients are obtained. That is, the quantization effective coefficient is restored. Further, according to the quantization table in FIG. 10, the quantized effective coefficient is inversely quantized to obtain an effective coefficient (AC coefficient) (step S62).

Like this

The zigzag scan is inversely performed on the AC coefficient obtained in step S52 and the DC coefficient obtained in step S52, and the inverse processing of the two-dimensional DCT is performed to reproduce the image information of the block (step S63). . Figure 1 shows the obtained DCT coefficients and image data. See Figure 3.

In step S64, it is determined whether or not compressed data has been processed for all blocks. If not completed, step 51 and subsequent steps are repeated for the next block data (step S65). After completion the process with all of the block unit performs an inverse block of the resulting block data, to reproduce the original image data (step S ⁶ 6). As described above, the compressed data can be decompressed.

In the above embodiment, 8 pixels vertically and 8 pixels horizontally are defined as one block, but 16 pixels vertically and 16 pixels horizontally may be defined as one block. Also, theoretically, any vertical n pixels and horizontal m pixels may be defined as one block. Further, the shape of the block may be a shape other than a rectangle. The shape of the blocks is preferably selected so that the data between the blocks has a correlation.

FIG. 21 shows a comparison between the compression method according to the present invention and the conventional compression method. The vertical axis is the SN ratio, that is, the signal-to-noise ratio · image quality, and the horizontal axis is the bit rate after compression and encoding (average number of encoded bits per pixel), which is inversely proportional to the compression ratio. Indicates the number of bits after compression for the original image of 8 bits per pixel. One block was 16 pixels vertically and 16 pixels horizontally. As the original image, zelda, Lena, and Mandrill, which are commonly used as evaluation images, were used. The present invention is shown by a solid line, and the conventional J PEG is shown by a broken line.

Compared to JPEG, it was found that the compression method of the present invention can achieve a compression ratio close to 1.5 to 2 times that of JPEG with the same playback image quality. It was also found that the SN ratio was improved by about 4 dB under the same compression ratio.

Figure 22 shows Lena used as the original image and its luminance histogram. FIG. 23 shows a Lena image compressed and decompressed according to the present invention and its luminance histogram. FIG. 24 shows a Lena image compressed and decompressed by the conventional JPEG and its luminance histogram (the compression ratio is the same as that of the present invention).

As clearly shown in FIGS. 22 to 24, it can be seen that the present invention has higher image quality after decompression. In the case of JPEG, many luminance components are lost in the luminance histogram, but in the present invention, the luminance component of the original image is well retained. ing.

FIG. 25 shows a compression device according to another embodiment, and FIG. 26 shows a decompression device. In this embodiment, the compression model is not selected, and all blocks are encoded. In the case of an image in which the maximum value XS1 exceeds 8 (or the effective effective coefficient length RElength exceeds 10) in most blocks, an embodiment as shown in FIGS. 25 and 26 is used. Is preferred. The operation of each unit is the same as in Figs.

In the above embodiments, the description has been made with respect to image data, but the present invention is also applicable to audio data, character text data, and the like.

Further, in each of the above embodiments, a case has been described in which each of the functions in FIGS. 1 and 2 is realized by using a CPU, but a part or all of the functions may be configured by hardware logic.

The compression / decompression method of the present invention can be widely applied to data recording, data communication, facsimile, digital camera, and the like.

Claims

The scope of the claims

1. Receiving a plurality of data given in order, outputting the value indicated by each data in accordance with a predetermined rule, outputting sorted value information, and giving the sorted order obtained by the above sorting and the given order. An alignment means for outputting the following reverse replacement information indicating the correspondence with the order;

Φ = ((Ϊ) Φ (2) φ (3) φ (Μ))

Here, ① is m that satisfies a (m) = j in the following replacement information σ,

σ = (σ (1) σ (2) σ (3) σ (Μ))

Here, σ φ indicates the number of the j-th data after sorting in the order before sorting, and M indicates the number of data.

PCR conversion means for receiving the reverse permutation information <»obtained by the alignment means and outputting the following PCR map m for the reverse permutation information,

m = [ms'ms IIIM] ^T

Here, the k-th element m _k is obtained by the following equation: th, Va (k) <a (k + l)

k-η ,,-Ι, Va (k)> HI (k + 1)

k = 2, 3, 4,, M where rik is the element of c) (a (k)) = k and the element of <> (a (k + l)) = k + l Is the number of elements whose value is less than k,

A PCR encoding unit that encodes the PCR map m obtained by the PCR encoding unit and outputs an encoded PCR map;

A data compression device based on permutation encoding, comprising:

. 2-Replacement Form (Rule 26)

2. The data compression device of claim 1,

Furthermore, orthogonal transform means for orthogonally transforming the given original data and providing the data as sorting target data to the sorting means is provided.

3. The data compression device of claim 2,

The orthogonal transformation means receives the original data, performs DCT transformation, obtains ZCT zigzag-scanned DCT transformation data, and gives the data as sorting target data to the sorting means.

4. The data compression device of claim 3, further comprising:

A quantization means for receiving the DCT conversion data and quantizing it based on a predetermined quantization roughness,

Effective coefficient extraction means for extracting, from the quantized DCT transformation data, data having the lowest frequency component to data having the highest frequency component exceeding a predetermined value as effective coefficients and providing the effective data to the sorting means as data to be sorted. ,

Characterized by having.

5. The data compression apparatus according to any one of claims 1 to 4,

The PCR encoding means is for Huffman encoding a PCR map.

6. The data compression device according to any one of claims 1 to 5,

The alignment value encoding means performs encoding by expressing the alignment value as a function.

7. The data compression device of claim 6,

The alignment value encoding means performs encoding by expressing a value obtained by multiplying an alignment value by an expansion coefficient as a function.

8. Receiving a plurality of data given in order, obtaining sorted value information in which the values indicated by each data are sorted according to a predetermined rule,

In order to show the correspondence between the sorted order obtained by the above sorting and the given order, order information in which the sorted order is arranged according to the given order is obtained, and the order information is arranged in a permutation. Obtaining structured order mapping information,

A data compression method characterized by encoding alignment value information and order mapping information, respectively, to obtain compressed data.

9. PCR decoding means for receiving the encoded PCR map in the given compressed data, decoding the encoded PCR map, and obtaining a PCR map;

A reverse PCR conversion means for receiving the PCR mapping obtained by the PCR decoding means and obtaining reverse replacement information corresponding to the PCR mapping,

Receiving an encoding alignment value information in the given compressed data, decoding the encoding alignment value information, and obtaining an alignment value information;

Restoring means for restoring original data based on the reverse replacement information obtained by the inverse PCR conversion means and the alignment value information obtained by the alignment value decoding means;

Data decompression device equipped with.

10. Decompose the given pressure box data to obtain the code mapping PCR mapping and the coding alignment value information,

Receiving the coded PCR map obtained by the above decomposition, decoding the coded PCR map to obtain a PCR map, obtaining the reverse permutation information corresponding to the PCR map, and obtaining the coding alignment obtained by the above decomposition. Decoding the value information to obtain alignment value information, and restoring the original data based on the reverse replacement information and the alignment value information.

11. An area for storing data relating to a DC component obtained by DCT conversion,

An area for storing coefficients of an approximate curve obtained by arranging AC components obtained by DCT transformation in an ascending order; An area for storing data on the following reverse replacement information indicating the correspondence between the order before sorting and the order after sorting,

A storage medium storing compressed data comprising:

Φ = (<> (1) Y (2) Y (3) Φ

Here, (j) is m that satisfies σ (ιη) ^ in the following replacement information σ,

a = (σ (1) σ (2) σ (3) σ (Μ))

Here, σ (D indicates the number of the j-th data after sorting in the order before sorting, M indicates the number of data,

1 2. A data transmission system comprising: (a) a transmitting device and (b) a receiving device,

(a) The transmitting device has the following (al) to (a6):

(al) Receives a plurality of data given in order, outputs the value indicated by each data according to a predetermined rule, outputs sorted value information, and gives the sorted order obtained by the above sorting and the given order. Sorting means for outputting the following permutation information indicating the correspondence with the order

Φ = (ψ (1) (2) ψ (3) Φ duck

Here, φ① is m that satisfies σ (m) = j in the following permutation information σ,

σ = (σ (1) σ (2) σ (3) σ (Μ))

Here, σ (j) indicates the order of the j-th data after sorting in the order before sorting, and M indicates the number of data.

(a2) receiving the inverse permutation information Φ obtained by the alignment means, and outputting the following PCR map m for the inverse permutation information Φ;

Here, the k-th element m _k is obtained by the following equation. ¾, Vo () <a (k + l)

k-iu-1, Va (k)>. (k + l)

k = 2, 3, M where iik is (where

Is the number of elements whose value is less than k between the element that is and () ( _CT (k + l)) = k + l.

(a3) a PCR encoding unit that encodes the PCR map m obtained by the PCR encoding unit and outputs an encoded PCR.

(b) The receiving device has the following (bl) to (! 35):

(bl) receiving means for receiving the transmitted compressed data,

(b2) a PCR decoding means for receiving an encoded PCR mapping in the received compressed data, decoding the encoded PCR mapping, and obtaining a PCR mapping,

(b3) inverse PCR conversion means for receiving the PCR map obtained by the PCR decoding means and obtaining reverse replacement information corresponding to the PCR map,

(b4) an alignment value decoding means for receiving the encoding alignment value information in the given compressed data, decoding the encoding alignment value information, and obtaining the alignment value information;

(b5) Restoring means for restoring the original data based on the reverse permutation information obtained by the inverse PCR conversion means and the alignment value information obtained by the alignment value decoding means.

1 3. The transmitting side compresses the data based on the data compression method of claim 8 and then transmits the data.

A data transmission method on the receiving side, wherein data is restored based on the data decompression method according to claim 10.

1 4. Receiving a plurality of data given in order, outputting the value indicated by each data according to a predetermined rule, and outputting sorted value information.

36-if i paper (Rule 26) Alignment means for outputting the following reverse replacement information indicating the correspondence between the obtained sorted order and the given order;

Φ = (Y (1) (2) (3) Φ (Μ)

Here, ① represents the following replacement information:

M that satisfies

Hi = (σ (1) σ (2) σ (3) hi (Μ))

PCR conversion means for receiving the inverse permutation information Φ obtained by the alignment means and outputting the following PCR map m for the inverse permutation information:

Here, the k-th element m _k is obtained by the following equation, iik, V. (k) ku a (k + l)

k-nk-1, Va (k)> hi (k + 1)

k = 2, 3,, M where nk is the element between <) (a (k)) = k and (a (k + l)) = k + l The number of elements whose value is less than k,

Storage means for storing the encoded PCR mapping from the PCR encoding means and the encoded alignment value information from the alignment direct encoding means as compressed data;

A data storage device using permutation encoding, comprising:

1 5. Receiving a plurality of data given in order, obtaining the sorted value information in which the value indicated by each data is sorted according to a predetermined rule,

In order to show the correspondence between the sorted order obtained by the above sorting and the given order, order information in which the sorted order is arranged according to the given order is obtained.

-37-Replacement Form (Rule 26) Obtaining order mapping information that structured the order information into a permutation,

Encode the alignment value information and the order mapping information, and obtain and store the compressed data

A data storage method characterized by the following.

16. A storage medium storing a program for realizing the device or method according to any one of claims 1 to 10 and 12 to 15 using a computer.

17. Compression model selecting means for receiving a plurality of data given in order and selecting whether to perform permutation coding or non-permutation coding based on the value indicated by the plurality of data,

When the permutation coding is selected by the compression model selection means, the compression by the following B (a) B coding means is executed, and when the non-permutation coding is selected by the compression model selection means. In (b), there is a compression execution means for executing compression by non-permutation encoding means,

(a) To obtain sorting value information in which values indicated by given data are sorted according to a predetermined rule, and to indicate a correspondence between the sorted order obtained by the sorting and the given order. , Obtains order information in which the order after sorting is arranged according to the given order, obtains order mapping information in which the order information is structured in a permutation, encodes and compresses the sort value information and the order mapping information, respectively. Permutation encoding means for obtaining data,

(b) When non-permutation encoding is selected by the compression model selection means, non-permutation encoding is performed, in which the given data is encoded to obtain compressed data without performing the above-described permutation encoding. Means,

A data compression device provided with.

18. The data compression apparatus according to claim 17,

Further, orthogonal transform means for orthogonally transforming the given original data and providing the result to the compression model selecting means is provided.

1 9. The data compression device according to claim 18,

The orthogonal transformation means receives the original data, performs DCT transformation, obtains ZCT zigzag-scanned DCT transformation data, and provides it to the compression model selection means.

20. The data compression apparatus according to claim 19,

Quantizing means for receiving the DCT transformed data from the orthogonal transform means and quantizing them according to a predetermined quantization roughness;

Among the DCT transform data quantized by the quantizing means, the data from the lowest frequency component and the highest frequency component exceeding a predetermined value are extracted as effective coefficient data, and the compression model selecting means Effective coefficient extracting means to give;

Characterized by having.

21. The data compression apparatus according to claim 20,

The compression model selection means includes, as a criterion for selecting a compression model, whether or not the maximum value of the AC component of the DCT conversion data is smaller than a predetermined value.

22. In the data compression apparatus of claim 20 or claim 21,

The compression model selecting means includes whether or not the number of the effective coefficients is smaller than a predetermined value as a criterion for selecting a compression box model.

23. The data compression apparatus of claim 17,

The replacement encoding means,

Receiving a plurality of data given in order, outputting the value indicated by each data in accordance with a predetermined rule, outputs sorted value information, and comparing the ordered order obtained by the above sorting with the given order. An alignment method that outputs the following installation information σ indicating the correspondence,

σ = (σ (1) σ (2) σ (3) σ (Μ)) Here, σ① indicates the number of the j-th data after sorting in the order before sorting, and M indicates the number of data.

Here, the k-th element m _k is obtained by the following equation:, Vff (k) <h (k + 1)

01k =

k-iik-1, VCT (k)> a (+ l)

k = 2,3,4,, M where _mk is the value in σ between the element with a (k) = k and the element with a (k + l) = k + l Is the number of elements less than k,

Characterized in that it is provided with:

24. The data compression device according to claim 17,

The replacement encoding means,

Receiving a plurality of data given in order, outputting the sort information in which the values indicated by each data are sorted according to a predetermined rule, and the correspondence between the sorted order obtained by the above sorting and the given order. Means for outputting the following reverse replacement information indicating

Φ = (Φ (ΐ) (2) (3) (Μ))

Here, yu (j) is m that satisfies a (m) = j in the following replacement information σ.

σ = (σ (1) σί2) σ (3) σ (Μ))

-40-Replacement sheet (Rule 26) Here, σ (j) indicates the number of the j-th data after sorting in the order before sorting, and M indicates the number of data.

PCR conversion means for receiving the reverse permutation information </> obtained by the alignment means and outputting the following PCR map m for the reverse permutation information <

m = [m :, m, m «] ^T

Here, the k-th element m _k is obtained by the following equation.

¾, Vo (k) <HI (k + 1)

Ok =

ki -l, V. (k)>. (k + l)

k = 2, 3, 4 M where is the value of between the element with 0 ( _CT (k)) = k and the element with φ (bi (k + l)) = c + l at Is less than k, the number of elements,

Characterized in that it is provided with:

25. The data compression apparatus according to any one of claims 23 to 24,

The alignment value encoding means is characterized in that, in the encoding of the alignment value, encoding is performed by expressing a value obtained by multiplying the alignment value by an expansion coefficient as a function.

26. The data compression apparatus according to any one of claims 17 to 25,

The non-permutation encoding means encodes data based on the JPEG standard.

27. Receiving a plurality of data given in order, determining whether to perform permutation coding or non-permutation coding based on the value indicated by the plurality of data, When the encoding is selected, the compression by the permutation encoding shown in (a) below is executed, and when the non-permutation encoding is selected by the compression model selecting means, the non-permutation encoding shown in (b) below is performed. A data compression method characterized in that compressed data is obtained by performing compression by

(a) When encoding is selected, the value indicated by the given data is sorted according to a predetermined rule to obtain sorting value information, and the sorted order obtained by the above sorting and the given order are given. In order to show the correspondence with the order, obtain order information in which the sorted order is arranged in accordance with the given order, obtain order place mapping information in which the order information is structured in order, and obtain the sort value. Information and order mapping information are encoded to obtain compressed data.

(b) If the non-permutation encoding is selected, the given data is encoded without performing the above-described substitution encoding to obtain compressed data.

28. A compression model judging means for judging whether compression processing by non-permutation encoding has been performed based on the compression box model data in the given compressed data,

If it is determined that the encoding is permutation encoding, the decompression is performed by the permutation code decompression means shown in (a) below. A data decompression device that performs decompression,

(a) If it is determined that the encoding is permutation encoding, an encoded PCR map is obtained from the given compressed data, the encoded PCR map is decrypted, and a PCR map is obtained. The corresponding reverse replacement information is obtained, the encoded alignment value information is obtained from the given compressed data, the encoded alignment value information is decoded, the alignment value information is obtained, and the reverse replacement information and the alignment value are obtained. Permutation decoding means for restoring data based on information

(b) Non-permutation decoding means for decoding the given compressed data and restoring the data without performing the above-described permutation decoding when determining that the encoding is non-permutation coding.

2 9. Based on compression model data being compressed data given, determines whether the compression processing has been made by the mosquitoes non location ^ ¹ No. of the compression process have been made by replacing the coding,

If it is determined that the encoding is permutation encoding, the following decompression is performed by the substitution symbol (a) below. A data decompression method characterized by executing

(a) If it is determined that the encoding is permutation encoding, an encoded PCR map is obtained from the given compressed data, the encoded PCR map is decrypted, and a PCR map is obtained. The corresponding reverse replacement information is obtained, the encoded alignment value information is obtained from the given compressed data, the encoded alignment value information is decoded, the alignment value information is obtained, and the reverse replacement information and the alignment value are obtained. Restore data based on information,

(b) If it is determined that the encoding is non-permutation encoding, the given compressed data is decoded to restore the data without performing the above-described permutation decoding.

30. A storage medium storing compressed data having the following structure:

A compression model storage area for storing a compression model;

When the compression model stored in the compression model storage area is the permutation coding, the area has the following areas (a) to (c),

(a) an area for storing data relating to a DC component by the DCT conversion,

(b) an area for storing a coefficient of an approximate curve obtained by arranging the AC components obtained by the DCT transformation in a substantially large order,

(c) An area for storing data indicating the correspondence between the order before sorting and the order after sorting If the compression model stored in the compression model / re storage area is non-replacement coding, the following (d) ~ (E) area,

(d) an area for storing data relating to the DC component due to the DCT transformation,

(e) An area that stores data on AC components obtained by DCT conversion without rearranging the order.

3 1. A data transmission system comprising (a) a transmitting device and (b) a receiving device fi as follows:

(a) The transmitting device has the following (al) to (a3):

(a2) When the replacement coding is selected by the compression model selecting means, the compression by the permutation coding means of (a21) below is executed, and the non-permutation coding is selected by the compression model selecting means. In this case, a compression executing means for executing the compression by the non-permutation encoding means (a22) below,

(a21) In order to obtain sorting value information in which the values indicated by the given data are sorted according to a predetermined rule, and to show the correspondence between the sorted order obtained by the above sorting and the given order, A permutation that obtains order information in which the order after sorting is arranged according to the order, obtains order mapping information in which the order information is structured in a permutation, and encodes the sorted value information and the order mapping information to obtain compressed data. Encoding means,

(a22) When the non-permutation encoding is selected by the compression model selection means, the non-permutation encoding is performed to obtain the compressed data by encoding the given data without performing the above-described permutation encoding. Encoding means,

(b) The receiving device has the following (bl) to (b3):

(bl) receiving means for receiving the transmitted compressed data,

(b2) compression model determining means for determining whether compression processing by permutation encoding has been performed based on the compression model data in the received compressed data, and whether compression processing by non-permutation encoding has been performed;

(b3) If it is determined that the encoding is the replacement encoding, the decompression by the replacement code decompression means described in (b31) below is executed. A data decompression device that performs decompression by code decompression means; (b31) If it is determined that the encoding is permutation encoding, an encoded PCR mapping is obtained from the given compressed data, and the encoded PCR mapping is decoded to obtain a PCR mapping. Obtain the corresponding reverse replacement information, obtain the encoding alignment value information from the given compressed data, decode the encoding alignment value information, obtain the alignment value information, and obtain the alignment replacement information and the alignment. Replacement decryption means for restoring data based on value information,

0> 32) Non-permutation decoding means for decoding the given compressed data and restoring the data without performing the above-described permutation decoding when determining that it is non-permutation encoding.

32. A compression model selecting means for receiving a plurality of data given in order and selecting whether to perform permutation coding or non-replacement coding based on the value indicated by the plurality of data,

If the replacement coding is selected by the compression model selection means, the following ( _a ) compression by the replacement coding means is executed, and if non-replacement coding is selected by the compression model selection means, (b) a means for performing compression by the non-permutation encoding means,

(a) To obtain sorting value information in which values indicated by given data are sorted according to a predetermined rule, and to indicate a correspondence between the sorted order obtained by the above sorting and the given order, In addition to obtaining order information in which the order after sorting is arranged in accordance with the specified order, obtaining order mapping information in which the order information is structured into a permutation, and signing the order value information and the order mapping information as code symbols. Permutation encoding means for obtaining compressed data

(b) When non-permutation encoding is selected by the compression model selection means, non-permutation is performed in which the given data is encoded to obtain compressed data without performing the above-described permutation encoding. Encoding means,

Storage means for storing the compressed data,

Data storage device equipped with

3 3. Receiving a plurality of data given in order, determining whether to perform permutation coding or non-permutation coding based on the value indicated by the plurality of data,

If the permutation encoding is selected, the compression by the permutation encoding shown in (a) below is executed, and if the non-permutation encoding is selected by the compression model selecting means, the following non-permutation encoding in (b) is performed. A data storage method comprising: performing compression by encoding to obtain compressed data; and storing the obtained compressed data.

(a) When the permutation coding is selected, the value indicated by the given data is sorted according to a predetermined rule to obtain sorting value information, and the order after the sorting obtained by the above sorting is given. In order to show the correspondence with the given order, in accordance with the given order, we obtain order information in which the sorted order is arranged, and obtain order place mapping information in which the order information is structured in order. Encoding the sorted value information and the order mapping information to obtain compressed data,

(b) When the non-permutation encoding is selected, the compressed data is obtained by encoding the given data without performing the above-described substitution encoding.

34. A storage medium storing a program for realizing the apparatus or method according to any one of claims 17 to 29 and 31 to 33 using a computer.