JP6913320B2 - Data processing equipment, methods and programs - Google Patents

Description

The present invention relates to a data processing apparatus, method and program for processing data.

A technique for generating digital data in which input information is converted into a predetermined character code by reading a document in which characters are written by hand with an image scanner or the like and performing OCR (Optical Character Recognition) processing is widespread.

For example, according to Patent Document 1, a character identification system that recognizes characters by machine learning is disclosed. The character recognition system has a character image input receiving unit that accepts input of a sample character image, character component extraction that extracts character parts based on the sample character image, and pseudo character model generation that generates a pseudo character model based on the character parts. A system consisting of a unit and an identification dictionary generation that generates a character identification pattern based on a pseudo-character model to generate an identification dictionary has been proposed.

Further, when character recognition is performed by a cloud server, it is conceivable to use a data compression technique in order to transmit the digital data read by the image scanner to the server via the network.

For example, according to Patent Document 2, a compressed data generation unit that compresses compression target data to generate compressed data and a deformed image that generates a plurality of types of deformed image data obtained by deforming the original image data with different amounts of deformation. The data generation unit, the difference calculation unit that calculates the difference data indicating the difference between each of the deformed image data and the original image data as the compression target data of the compression data generation unit, and each of the compressed data of each difference data are deformed. An image processing device including a control unit that stores in a compressed data storage unit in association with the amount of deformation in the image data generation unit has been proposed.

Japanese Unexamined Patent Publication No. 2015-069256 Japanese Unexamined Patent Publication No. 2008-236202

However, although Patent Document 2 discloses an example of using a Huffman table for run-length coding, there is a desire to further increase the compression ratio.

An object of the present invention is to provide a data processing apparatus, method and program capable of converting a document containing various handwritten characters into digital data having a high compression ratio.

In order to achieve the above object, the data processing device according to one aspect of the present invention includes a rotation processing unit that rotates the image data and a compression unit that compresses the image data rotated by the rotation processing unit. In the image data compression process, the compression unit converts the image data into binarized data having the same continuous value, and encodes the converted binarized data.

Further, the data processing method according to one aspect of the present invention includes a rotation processing step of rotating the image data by a predetermined angle by the rotation processing unit and a compression process of compressing the image data rotated by the rotation processing unit by the compression unit. The compression step includes a step, and in the image data compression process, the image data is converted into binarized data in which the same value is continuous, and the converted binarized data is encoded.

Further, the data processing program according to one aspect of the present invention is for realizing a rotation processing step of rotating the image data by a predetermined angle and a compression process of compressing the image data rotated by the rotation processing step by a computer. In the image data compression process, the data processing program is a program that converts image data into binarized data in which the same value is continuous, and encodes the converted binarized data. be.

According to the present invention, a document containing various handwritten characters can be converted into digital data having a high compression rate.

It is a block diagram which shows the structure of a data processing apparatus. It is a figure which provides the explanation of the procedure of extracting the item image data from the image data by the extraction unit. It is a figure which provides the explanation about the operation of the compression part. FIG. 1 is a first diagram for explaining the relationship between the histogram generated from the binarized data of the item image data and the rotation angle of the item image data. FIG. 2 is a second diagram for explaining the relationship between the histogram generated from the binarized data of the item image data and the rotation angle of the item image data. It is a figure which provides the explanation about the procedure of determining a rotation angle. It is a flowchart which provides the explanation about the operation of the 1st processing part. It is a flowchart which provides the explanation about the operation of the 2nd processing part.

Hereinafter, the data processing apparatus, method, and program according to the embodiment of the present invention will be described with reference to the drawings. In all the drawings for explaining the embodiments, the common components are designated by the same reference numerals, and the repeated description will be omitted.

In the following, as an example, the configuration and operation of a data processing device that converts documents such as forms and questionnaires containing handwritten characters into image data with a scanner or the like and compresses the converted image data will be described. It should be noted that a document that does not include handwritten characters, that is, a document in which characters are printed by a printer using typeface data, or the like may be targeted.

As shown in FIG. 1, the data processing device 1 includes an image data conversion unit 10, an extraction unit 11, a rotation processing unit 12, a compression unit 13, and an output unit 14.

The image data conversion unit 10 converts a document composed of a plurality of items into image data. For example, the image data conversion unit 10 is configured by a scanner device, images a document to generate image data, and inputs the generated image data to the image data storage unit 21.

The extraction unit 11 reads the image data converted by the image data conversion unit 10 from the image data storage unit 21 and extracts the image data in item units. Specifically, the extraction unit 11 performs document structure analysis such as ruled line extraction, frame structure analysis, and position estimation of the reading target frame based on the image data, and extracts the image data in item units. FIG. 2 is an example schematically showing how four data (item data a1 to a4 in FIG. 2) are extracted from the image data A by the extraction unit 11. The extraction unit 11 extracts the item image data a1 from the image data A as a katakana item, extracts the item image data a2 from the image data A as a name item, and extracts the item image data a3 from the image data A as an address item. The item image data a4 is extracted from the image data A as a telephone item.

The rotation processing unit 12 rotates the image data extracted by the extraction unit 11. Specifically, the rotation processing unit 12 is composed of a first processing unit 15 and a second processing unit 16.
The first processing unit 15 rotates the image data at a predetermined angle. When the image data is composed of a plurality of items, the first processing unit 15 rotates the item image data extracted by the extraction unit 11.

According to the first processing unit 15, the image data is rotated at a predetermined angle, and then the compression processing is performed by the compression unit 13. Therefore, it is possible to increase the series processing speed while improving the compression rate. can. Further, although the details will be described later, according to the second processing unit 16, the compression rate of the image data can be further improved.

For example, the data processing device 1 may be configured such that the first processing unit 15 initially performs rotation processing of the image data, and when the compression rate is equal to or less than a predetermined value, the second processing unit 16 rotates the image data. .. Further, the data processing device 1 rotates the image data by the first processing unit 15 when the processing speed is prioritized, and rotates the image data by the second processing unit 16 when it is desired to increase the compression rate. It may be configured to be processed.

<About the configuration and operation of the rotation processing unit>
As shown in FIG. 1, the second processing unit 16 includes a binarized data conversion unit 17, a statistical processing unit 18, an image data rotation unit 19, and a rotation angle determination unit 20.

The binarized data conversion unit 17 converts the image data into the binarized data. When the image data is composed of a plurality of items, the binarized data conversion unit 17 may be configured to convert the item image data extracted by the extraction unit 11 into the binarized data. In the following, processing of item image data will be described as an example. In the item image data, the background and the characters are separated by binarization. The item image data is a shade image composed of a background and characters. The binarized data conversion unit 17 separates the background and characters by converting the item image data into data consisting of two types of density values, black (for example, “1”) and white (for example, “0”). do. In this embodiment, black is set to "1" and white is set to "0", but black may be set to "0" and white may be set to "1".

The statistical processing unit 18 performs statistical processing on the binarized data converted by the binarized data conversion unit 17. Statistical processing is, for example, generating a histogram, which will be described in detail later.

The image data rotation unit 19 rotates the item image data by a predetermined angle after performing statistical processing by the statistical processing unit 18. The predetermined angle is, for example, 0.1 degrees.

The rotation angle determination unit 20 performs processing by the binarization data conversion unit 17 and the statistical processing unit 18 every time the item image data is rotated by the image data rotation unit 19, and is based on the comparison of the statistical processing results for each rotation angle. , Determine the rotation angle of the item image data.

The compression unit 13 rotates the item image data at a rotation angle determined by the rotation angle determination unit 20, and performs compression processing of the rotated item image data.

Further, in the compression process of the item image data, the compression unit 13 converts the item image data into binarized data in which the same value is continuous, and encodes the converted binarized data. Although the details will be described later, the item image data is compressed at an angle in which black and white are continuous. Therefore, the binarized data is data in which the same value is continuous, that is, a plurality of "1" (for example, black) are continuous, and a plurality of "0" (for example, white) are continuous. It has become.

Further, as the coding, for example, run-length coding or the like can be considered, but other coding may be used. Further, the compression unit 13 may be configured to convert the encoded data based on the conversion table.

<About the operation of the compression unit>
Here, the operation of the compression unit 13 will be described with reference to FIG. In the following, as shown in FIG. 3A, the binarized data has five consecutive whites (that is, "0"), then five consecutive blacks (that is, "1"), and then. An example is the case where five white colors (that is, "0") are consecutive.

The compression unit 13 encodes "000000111100000" by run-length coding. The encoded data becomes "W5B5W5" as shown in FIG. 3 (b). That is, at this point, the 15-bit information can be compressed to 6 bits. Further, the compression unit 13 converts the encoded data based on the conversion table as shown in FIG. 3D. Specifically, "W5" can be converted to "E" based on FIG. 3 (d), and "B5" can be converted to "e" based on FIG. 3 (d). Therefore, the converted data becomes "EeE" as shown in FIG. 3 (c). Therefore, according to this embodiment, the 15-bit binarized data can be compressed (compressed to 1/5) into 3-bit data.

In this way, documents centered on character information often have a series of blacks and a series of whites, and the compression method as described above is effective. Further, as will be described later, a remarkable effect can be obtained by performing pre-processing (rotation processing of item image data) so that the continuity of black and white is further increased.

<Compressed data>
A symbol indicating an item may be inserted into the data compressed by the compression unit 13. For example, the symbol for katakana items is "$ 1", the symbol for name items is "$ 2", the symbol for address items is "$ 3", and the symbol for telephone items is "$ 3". It is assumed to be "$ 4". The compression unit 13 inserts a symbol of a corresponding item while compressing the data.

For example, the compression unit 13 compresses data such as "FtVe ... Rt $ 1jTpX ... YuOp ... W $ 2gHnM ... uR $ 3WqVm ... Pg $ 4". On the receiving side that can support the compression method, "FtVe ... Rt" is a katakana item ($ 1), and "jTpX ... YuOp ... W" is a name item ($ 2). It can be seen that "gHnM ... uR" is an address item ($ 3) and "WqVm ... Pg" is a telephone item ($ 4). In this way, the data processing device 1 can further compress and encrypt the information, and can improve the security of the compressed data.

<Relationship between histogram and rotation angle of item image data>
The statistical processing unit 18 generates a histogram of the binarized data converted by the binarized data conversion unit 17.

Here, the relationship between the histogram generated from the binarized data of the item image data of "two", which is a handwritten Chinese numeral, and the rotation angle of the item image data will be described with reference to FIG. The direction in which the histogram is generated is the X direction as shown in FIG. 4A. For example, in the binarized data generated from the item image data, black is integrated for each row, and a histogram is generated by graphing based on the integrated value.

When the item image data is arranged in the X direction as shown in FIG. 4A, the statistical processing unit 18 integrates black for each row as shown in FIG. 4B. Generate a histogram based on the integrated values. When the item image data is arranged in the X direction as shown in FIG. 4 (c), the statistical processing unit 18 integrates black for each row as shown in FIG. 4 (d). Then, a histogram is generated based on the integrated value. When the item image data is arranged in the X direction as shown in FIG. 4 (e), the statistical processing unit 18 integrates black for each row as shown in FIG. 4 (f). Then, a histogram is generated based on the integrated value.

Further, the rotation angle determination unit 20 determines the angle at which the peak of the histogram becomes the maximum or the maximum as the rotation angle of the item image data. Comparing FIGS. 4 (b), (d), and (f), the histogram pattern shown in FIG. 4 (b) has the most continuous white and continuous black. Therefore, the rotation angle determination unit 20 determines the rotation angle so that the item image data has the angle shown in FIG. 4A.

Here, the relationship between the orientation of characters written on a document and the compression ratio will be described. In the rotation correction that is usually performed, the document is corrected so that it faces sideways. However, according to the method according to the present embodiment, the rotation correction is performed so that the orientation of the characters written on the document, not the orientation of the document, is right beside the writer in consideration of the habit of the writer. This increases the number of consecutive blacks and whites, and the compression process described above further improves the compression ratio.

<Direction to create a histogram>
Further, it is preferable that the statistical processing unit 18 creates a histogram in a direction orthogonal to the description direction of the characters included in the item image data. For example, as shown in FIG. 5A, the writing direction of the characters in the document is the direction in which the vertical directions are orthogonal to each other as shown in FIG. 5B in the case of horizontal writing. In addition, when the writing direction of characters in a document is vertical writing, the horizontal direction is orthogonal.

Here, the relationship between the histogram generated from the binarized data of the item image data of "patent", which is a handwritten kanji, and the rotation angle of the item image data will be described with reference to FIG.

When the item image data is arranged as shown in FIG. 5 (a) (direction parallel to the writing direction of the characters), the statistical processing unit 18 performs line by line as shown in FIG. 5 (b). Black is integrated and a histogram is generated based on the integrated value. Further, when the item image data is arranged as shown in FIG. 5 (c) (direction orthogonal to the writing direction of the characters), the statistical processing unit 18 performs a line as shown in FIG. 5 (d). Black is integrated for each, and a histogram is generated based on the integrated value.

The white portion in FIG. 5 (a) is “A1 + A2”, and the white portion in FIG. 5 (b) is “B1 + B2 + B3”. Therefore, it can be seen in FIG. 5 (b) that there are many continuous white portions. That is, when compressing item image data of the same size, as shown in FIG. 5C, it is better to create a histogram in a direction orthogonal to the writing direction of the characters included in the item image data. The rate is high.

<Procedure for determining the rotation angle>
Next, the procedure for determining the rotation angle will be described with reference to FIG. As described above, the image data rotation unit 19 rotates the item image data by 0.1 degrees after performing statistical processing by, for example, the statistical processing unit 18.

The binarized data conversion unit 17 converts the rotated item image data into binarized data. Next, the statistical processing unit 18 generates a histogram of the binarized data converted by the binarized data conversion unit 17.

After that, the rotation process by the image data rotation unit 19, the binarization process by the binarization data conversion unit 17, and the generation of the histogram of the binarized data by the statistical processing unit 18 are repeated.

As shown in FIG. 6, the rotation angle determining unit 20 compares the histograms for each rotation angle and determines the angle at which the peak of the histogram becomes the maximum or the maximum as the rotation angle of the item image data. In the example shown in FIG. 6, the rotation angle of X1 is determined.

<First method>
Next, the procedure for compressing the image data when the first processing unit 15 is used will be described with reference to the flowchart shown in FIG. 7.

In step S1, the rotation processing unit 12 rotates the image data at a predetermined angle.

In step S2, the compression unit 13 compresses the image data rotated by the rotation processing unit 12.

Therefore, the data processing device 1 uses the first processing unit 15 to increase the processing speed of a document containing various handwritten characters while compressing the image data with a high compression rate. be able to.

<Second method>
Next, the procedure for compressing the image data when the second processing unit 16 is used will be described with reference to the flowchart shown in FIG.

In step S11, the binarized data conversion unit 17 converts the image data into the binarized data.

In step S12, the statistical processing unit 18 performs statistical processing on the binarized data converted by the binarized data conversion unit 17.

In step S13, the image data rotation unit 19 rotates the image data by a predetermined angle after performing statistical processing by the statistical processing unit 18.

In step S14, the rotation angle determination unit 20 determines whether or not to determine (determine) the rotation angle of the image data. If the rotation angle is determined, the process proceeds to step S15, and if the rotation angle is not determined, the process returns to step S11. Specifically, the rotation angle determination unit 20 performs processing by the binarization data conversion unit 17 and the statistical processing unit 18 every time the image data is rotated by the image data rotation unit 19, and the statistical processing result for each rotation angle is obtained. Based on the comparison, the rotation angle of the image data is determined.

In step S15, the compression unit 13 rotates the image data at a rotation angle determined by the rotation angle determination unit 20, and performs compression processing of the rotated image data.

Therefore, the data processing device 1 can convert a document containing various handwritten characters into digital data having a higher compression rate by using the second processing unit 16.

<Program>
Further, in this embodiment, the configuration and operation of the data processing device 1 for converting a document containing various handwritten characters into digital data having a high compression rate have been mainly described, but the present invention is not limited to this, and each configuration is not limited to this. It may be configured as a method and program for converting a document having elements and containing various handwritten characters into digital data with a high compression ratio.

Further, it is realized by recording a program for realizing each function constituting the data processing device 1 on a computer-readable recording medium, reading the program recorded on the recording medium into a computer system, and executing the program. You may.

Specifically, the program is a data processing program for realizing a rotation processing step of rotating image data by a predetermined angle and a compression process of compressing image data rotated by the rotation processing step by a computer. be. Further, in the compression step, in the image data compression process, the image data is converted into binarized data in which the same value is continuous, and the converted binarized data is encoded.

Further, the term "computer system" as used herein includes hardware such as an OS and peripheral devices. Further, the "computer-readable recording medium" refers to a portable medium such as a flexible disk, a magneto-optical disk, a ROM, or a CD-ROM, or a storage device such as a hard disk built in a computer system.

Furthermore, a "computer-readable recording medium" is a medium that dynamically holds a program in a short period of time, such as a communication line when a program is transmitted via a network such as the Internet or a communication line such as a telephone line. , In that case, a program that holds a program for a certain period of time, such as a volatile memory inside a computer system that serves as a server or a client, may be included. Further, the above-mentioned program may be a program for realizing a part of the above-mentioned functions, and may be a program for realizing the above-mentioned functions in combination with a program already recorded in the computer system. ..

1 data processing device, 10 image data conversion unit, 11 extraction unit, 12 rotation processing unit, 13 compression unit, 14 output unit, 15 first processing unit, 16 second processing unit, 17 binarized data conversion unit, 18 statistics Processing unit, 19 image data rotation unit, 20 rotation angle determination unit, 21 image data storage unit, 22 compressed data storage unit

Claims (8)

A rotation processing unit that rotates image data,
A compression unit that performs compression processing of image data rotated by the rotation processing unit is provided.
The rotation processing unit performs statistical processing on the binarized data obtained by converting the image data into binarized data, compares the statistical processing results each time the image data is rotated by a predetermined angle, and rotates the image data. Decide and
In the image data compression process, the compression unit rotates the image data at the determined rotation angle , converts the rotated image data into binarized data in which the same value is continuous, and the converted binary data. A data processing device that encodes compressed data. The rotation processing unit is composed of a first processing unit and a second processing unit.
The first processing unit rotates the image data at a predetermined angle to rotate the image data.
The second processing unit
A binarized data conversion unit that converts image data into binarized data,
A statistical processing unit that performs statistical processing on the binarized data converted by the binarized data conversion unit, and
An image data rotating unit that rotates the image data by a predetermined angle after performing statistical processing by the statistical processing unit.
The processing by the binarization data conversion unit and the statistical processing unit is performed every time the image data is rotated by the image data rotation unit, and the rotation angle of the image data is determined based on the comparison of the statistical processing results for each rotation angle. Equipped with a rotation angle determination unit
The data processing device according to claim 1, wherein the compression unit rotates image data at a rotation angle determined by the rotation angle determination unit, and performs compression processing of the rotated image data. The statistical processing unit generates a histogram of the binarized data converted by the binarized data conversion unit.
The data processing device according to claim 2, wherein the rotation angle determining unit determines an angle at which the peak of the histogram becomes the maximum or the maximum as the rotation angle of the image data. The data processing apparatus according to claim 2 or 3, wherein the statistical processing unit creates a histogram in a direction orthogonal to the writing direction of characters included in the image data. An image data conversion unit that converts a manuscript composed of multiple items into image data,
Equipped with an extraction unit that extracts image data item by item
The data processing device according to any one of claims 2 to 4, wherein the extraction unit extracts image data converted by the image data conversion unit on an item-by-item basis. The data processing device according to claim 1, wherein the compression unit converts encoded data based on a conversion table. A rotation processing process in which the image data is rotated by a predetermined angle by the rotation processing unit, and
The compression unit includes a compression step of performing compression processing of the image data rotated by the rotation processing unit.
The rotation processing unit performs statistical processing on the binarized data obtained by converting the image data into binarized data, compares the statistical processing results each time the image data is rotated by a predetermined angle, and rotates the image data. Decide and
In the compression step, in the image data compression process, the image data is rotated at the determined rotation angle, the rotated image data is converted into binarized data in which the same value is continuous, and the converted binary data is converted. A data processing method that encodes compressed data. A rotation processing process that rotates the image data by a predetermined angle,
A data processing program for realizing a compression process of compressing image data rotated by the rotation processing process by a computer.
In the rotation processing step, statistical processing is performed on the binarized data obtained by converting the image data into binarized data, the statistical processing results are compared each time the image data is rotated by a predetermined angle, and the rotation angle of the image data is compared. Decide and
In the compression step, in the image data compression process, the image data is rotated at the determined rotation angle, the rotated image data is converted into binarized data in which the same value is continuous, and the converted binary data is converted. A data processing program that encodes compressed data.

Images