JP3373068B2 - Optical character recognition device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an optical character recognition device for reading characters from a general document image pattern of newspapers, magazines, etc. and converting them into electronic files.

[0002]

2. Description of the Related Art Conventionally, an optical character recognizing device sets an information medium of a general document such as a newspaper or a magazine in an image acquiring device such as an image scanner and starts reading the information medium by an image reading means. The document image pattern is optically read and stored in the document image pattern storage means, and the document image pattern is subjected to preprocessing such as character / figure separation, line segmentation, and character segmentation, and character recognition process, and character recognition is performed. Graphic information (consisting of characters, figures, positions, attributes, etc.) is stored in the character / graphic information storage means.

[0003]

In the conventional optical character recognition device, an information medium of a general document (hereinafter referred to as a book type) having a two-page spread like a magazine is set in the image acquisition device. When the recognition processing is performed, even if the desired information is the page unit of the magazine, the information for two pages is recognized and stored as the information for one page. Therefore, in order to store the information for each page of the magazine, The stored information has to be manually edited again, which is troublesome.

Alternatively, when the information medium is set in the image acquisition device, there is also a method of setting the page in the image acquisition device to acquire the image, but there is a problem that it takes time in any case.

It is an object of the present invention to provide an optical character recognition device capable of acquiring information for a plurality of pages as an image and storing character graphic information in page units.

[0006]

In order to achieve the above object, in the optical character recognition apparatus of the present invention, a value indicating the area range of the document image pattern is input from the document image pattern storage means, The page division information processing means for processing and calculating the page division information and the document image pattern from the document image pattern storage means are input, and the document image pattern is subjected to preprocessing and character recognition processing, and page division is performed. And a page unit character recognition processing means for outputting character and graphic information to the character and graphic information storage means on a page basis based on the information.

[0007]

The page-based character recognition processing means of the optical character recognition apparatus configured as described above performs preprocessing and character recognition processing on the document image pattern, and performs page division information processing on the recognized character graphic information. The data is output to the character / graphic information storage means in page units based on the page division information input from the means.

Therefore, according to the present invention, information of a plurality of pages can be obtained as an image and the character / graphic information can be stored in page units.

[0009]

Embodiments of the present invention will be described with reference to the drawings. Elements common to the drawings are given the same reference numerals.

First Embodiment FIG. 1 is a functional block diagram showing the basic configuration of the present invention. The image reading means 1 optically reads the information medium and outputs it to the document image pattern storage means 2. The page division information processing means 3 processes a value indicating the area range of the document image pattern input from the document image pattern storage means 2 to calculate page division information. The page unit character recognition processing means 4 inputs the document image pattern from the document image pattern storage means 2 and performs preprocessing such as character pattern separation, line segmentation, and character segmentation on the document image pattern and character recognition process. The applied and recognized character / graphic information is output to the character / graphic information storage means 5 in page units based on the page division information input from the page division information processing means 3.

FIG. 2 is a block diagram showing the configuration of the first embodiment. The central processing unit 11 (hereinafter referred to as CPU 11) includes an image memory 12 and a main memory 13 (hereinafter referred to as the memory 13).
The image acquisition device 14 such as an image scanner is connected to the input section of the image memory 12. The central processing device 11 and the image acquisition device 14 constitute the image reading means 1, the central processing device 11 and the image memory 12 constitute the document image pattern storage means 2, and the central processing device 11 and the memory 13 constitute the document image pattern storage means 2. The page division information processing means 3 and the page unit character recognition processing means 4 are configured.

FIG. 3 is a schematic diagram of the document image pattern of the first embodiment. In the document image pattern, the CPU 11 outputs a read timing signal to the image memory 12 and the image acquisition device 14, and the image acquisition device 14 optically reads the information medium and stores it in the image memory 12. The document image pattern has an area range of width W in the positive direction of the X axis and height H in the negative direction of the Y axis from the origin 0. If the information medium is a book or the like, it is between the document image areas 21 and 22. There is a shaded portion 20.

FIG. 4 is a histogram of the document image pattern shown in FIG. Horizontal axis is X, vertical axis is histogram value H
(X), the Y-axis of the document image pattern shown in FIG. 3 is scanned in the positive direction of the X-axis at a predetermined interval, and the frequency distribution of black data is represented as a histogram value H (X). Is.
Since the document image areas 21 and 22 have a small frequency distribution of black data, they are below a predetermined threshold value H (s), and the shaded area 20 has a large frequency distribution of black data, and therefore exceeds the threshold value H (s). .

FIG. 5 is a flow chart for explaining the operation of the first embodiment.
It is a chart. In step S1, the CPU 11 outputs a read timing signal to the image memory 12 and the image acquisition device 14 to store the information optically read by the image acquisition device 14 in the image memory 12 as a document image pattern.

In step S2, the CPU 11 causes the image memory 1
The values W and H indicating the area range of the document image pattern shown in FIGS. 2 to 3 are input and stored in the memory 13.

In step S3, the CPU 11 scans the document image pattern to obtain a histogram and stores it in the memory 13, as shown in FIG.

In step S4, the CPU 11 sets the threshold value H (s).
The number of shaded portions 20 exceeding the number is counted to calculate the number of divided pages. When the number of divided pages is P, it is represented by P = the number of shaded portions + 1, and in this embodiment, since there is one shaded portion 20, the number of divided pages P = 2.

In step S5, the CPU 11 reads out the value W indicating the area range of the document image pattern stored in the memory 13, and divides it by the number of divided pages P as shown in FIG. Find B (Xk, 0). In this embodiment, X
Since k = W / P = W / 2, provisional page division position B (W /
2, 0).

In step S6, the CPU 11 uses the histogram to generate a temporary page division position B (Xk as shown in FIG.
, 0) to obtain the true page division position S (Xs, 0).
That is, the CPU 11 has preset ranges of δ and δ in the + direction and − direction centering on the temporary page division position Xk in the X-axis direction, and the histogram value H (X) is within the range.
The value Xs of X that maximizes the
s, 0) and the start and end positions of each page are stored in the memory 1
Store in 3. In this embodiment, the start position and end position of the first page are (0, 0) and (Xs, H), and the start position and end position of the second page are (Xs, 0) and (W, respectively).
H) is stored in the memory 13.

In step S7, the CPU 11 causes the image memory 1
The document image pattern is read based on the start position and end position of each page from 2, and pre-processing such as character and figure separation processing, line cutting processing, and character cutting processing is performed, and the cut-out processed characters are further processed. The recognition process is performed on the page, and the page is stored in the memory 13 page by page.

In step S8, the CPU 11 checks whether or not all divided pages have been processed. If all divided pages have been processed, the processing ends, and if no, the process returns to step S7.

In this embodiment, the operation in which the image acquisition device optically reads a book type information medium and divides the document image pattern stored in the image memory into pages is described. When this operation is performed over a period of time, steps S1 to S8 are repeated for a plurality of pages. At this time, the page number 1 given to the first divided page by the CPU 11 is stored in the memory 13, the page number 1 is given to the first divided page, and then each page is sequentially incremented by +1. Alternatively, the page number may be automatically assigned.

Further, although preprocessing and recognition processing are performed for each page unit in steps S7 and S8 based on the area stored in the memory 13 in step S6, the entire document image pattern is composed of characters and figures. Separation processing is performed, and then, based on the start position and end position of each page stored in the memory 13, pre-processing such as character / figure separation processing, line cutout processing, and character cutout processing is performed. The recognition processing may be performed and stored in the memory 13 page by page.

Further, after the preprocessing and recognition processing are performed on the entire document image pattern, it is stored in the memory 13 page by page based on the start position and end position of each page stored in the memory 13. Good.

According to the first embodiment, from the shaded area to the number of divided pages,
Since the page division position is automatically calculated, even a person who is unfamiliar with the operation can easily operate it.

Second Embodiment FIG. 6 is a block diagram showing the configuration of the second embodiment.
The difference from the embodiment is that the input device 15 is provided. The input device 15 is, for example, a key switch, and inputs the number of divided pages, the page addition direction, the first added page number, and the total number of added pages. The page addition direction is a parameter for deciding whether to add a page from the left to the right or a page from the right to the left after dividing the document image pattern stored in the image memory into pages. It is In the first embodiment, the number of divided pages is automatically obtained by the CPU 11 from the shaded area existing in the document image pattern, but in the second embodiment, it is designated by the input device 15.

FIG. 7 is a schematic diagram of the document image pattern of the second embodiment. In the document image pattern, the CPU 11 outputs a read timing signal to the image memory 12 and the image acquisition device 14, and the image acquisition device 14 optically reads the information medium and stores it in the image memory 12. The document image pattern has a range of a width W from the origin 0 in the positive direction of the X axis and a height H in the negative direction of the Y axis. Unlike the first embodiment, there is no shaded area.

FIG. 8 is a histogram of the document image pattern shown in FIG. Horizontal axis is X, vertical axis is histogram value H
(X), the Y-axis of the document image pattern shown in FIG. 7 is scanned in the positive direction of the X-axis at a predetermined interval, and the frequency distribution of black data is represented as a histogram value H (X). Is.

FIG. 9 is a flow chart for explaining the operation of the second embodiment.
It is a chart. The number of divided pages, the page addition direction, the first added page number, and the total number of added pages are input from the input device 15 and stored in the memory 13. In this embodiment, the number of divided pages P =
2. As for the page giving direction, pages are given from left to right.

In step S1, the CPU 11 sets the first information medium set in the image acquisition device 14 to the image memory 12,
A read timing signal is output to the image acquisition device 14, and the information optically read by the image acquisition device 14 is stored in the image memory 12 as a document image pattern.

In step S2, the CPU 11 causes the image memory 1
The values W and H indicating the area range of the document image pattern shown in FIGS. 2 to 7 are input and stored in the memory 13.

In step S3, the CPU 11 scans the document image pattern, obtains a histogram and stores it in the memory 13, as shown in FIG.

In step S4, the CPU 11 obtains a temporary page division position. That is, the number of divided pages P = 2 stored in the memory 13 and the value W indicating the area range of the document image pattern are read and divided to obtain a temporary page division position B (Xk, 0). In this embodiment, Xk = W / P = W / 2, so that the temporary page division position B (W / 2, 0) is obtained.

In step S5, the CPU 11 uses the histogram to generate a temporary page division position B (Xk
, 0) to obtain the true page division position S (Xs, 0).
That is, the CPU 11 moves around the provisional page division position Xk in the X-axis direction in the + direction and the − direction at predetermined intervals δ, and the histogram at that position (Xk ± N × δ: N is an integer). Value H (Xk ± N × δ) and predetermined threshold H (s)
And the position (Xk + N1 ×
.delta.: N1 is an integer) and (Xk-N2.times..delta.: N2 is an integer), and Xs = Xk + {(N1-N2) .times..delta.} /
2 is obtained and the true page division position S (Xs, 0) is set, and the start position and end position of each page are stored in the memory 13. In this embodiment, the start position and end position of the first page are (0,
0) and (Xs, H), and the start and end positions of the second page are (Xs, 0) and (W, H), respectively.
Remember.

In step S6, the CPU 11 causes the image memory 1
From 2, the document image pattern is read out based on the start position, end position of each page, the page giving direction, and the given page number, and pre-processing such as character and figure separation processing, line cutting processing, and character cutting processing is performed. Further, recognition processing is performed on the characters that have been cut out, the added page numbers are added, and the memory 13 is added in page units.
Remember. The CPU 11 assigns the first assigned page number to the first divided page, increments it by +1 to assign it to the next divided page, and subtracts the page number from the total number of assigned pages.

In step S7, the CPU 11 checks whether or not all the information media have been processed, depending on whether or not the total number of pages to be added has become 0. If all the processing has been completed, the process ends.

It should be noted that although pre-processing and recognition processing have been carried out page by page in steps S6 and S7 based on the area stored in the memory 13 in step S5, the entire document image pattern has characters and figures. Separation processing is performed, and then memory 1
Preprocessing such as line cutout processing and character cutout processing is performed based on the start position and end position of each page stored in No. 3, and recognition processing is performed on each cut out character, and stored in the memory 13 in page units. You may do it.

Further, after the preprocessing and recognition processing are performed on the entire document image pattern, it is stored in the memory 13 page by page based on the start position and end position of each page stored in the memory 13. Good.

Further, in this embodiment, the temporary page division position is obtained, and the true page division position is obtained by utilizing the black background portion of the histogram shown in FIG. 8. However, conversely, in the widest range of the white background. The position of the point may be the true page division position.

Although the page division position is corrected in this embodiment, the page division position is (W / P, 0) when the correction is not required.

Further, in this embodiment, the number of divided pages, the page giving direction, the first giving page number, and the total giving page number are inputted from the input device, but the page giving direction and the first giving page number are automatically set. If so, only the number of divided pages is required. Also,
Make sure to set the first page number automatically,
When pages are added in accordance with the page giving direction of the information medium, it is only necessary to input the number of divided pages and the page giving direction.

Although the number of divided pages is input in this embodiment, the divided page size may be input and the values W and H indicating the area range of the document image pattern may be divided to obtain the page division position. . When inputting the divided page size, the paper sizes A5, A4 and A3 are stored in the memory in accordance with the print paper size, and the input is A5, A4 and A3.

According to the second embodiment, the page division is performed by correcting the page division position even if the information medium is set in the image acquisition device with some deviation.

[0044]

Since the present invention is configured as described above, it has the following effects.

Input from the document image pattern storage means and page division information processing means for processing a value indicating the area range of the document image pattern input from the document image pattern storage means and outputting page division information. A page unit character recognition processing unit for performing preprocessing and character recognition processing on the formed document image pattern and outputting character graphic information in page units based on page division information input from the page division information processing unit, and page unit character recognition By providing the page unit information storage unit for inputting and storing character graphic information from the processing unit, the page unit character recognition processing unit performs preprocessing and character recognition processing on the document image pattern, and the recognition processing is performed. Since the character / graphic information is output to the page unit storage unit in page units based on the page division information input from the page division information processing unit, information for a plurality of pages can be acquired as an image and the character / graphic information can be stored in page units.

Further, the page division information processing means inputs the document image pattern from the document image pattern storage means, calculates the number of divided pages from the shaded portion between the pages of the document image pattern,
Further, the value indicating the area range of the document image pattern is divided by the number of divided pages to calculate the page division position, and is output as page division information to the page unit character recognition processing means.
Since the number of pages to be divided and the page division position are automatically calculated, even a person unfamiliar with the operation can easily perform the operation.

Further, it has a page division information input means for inputting the number of divided pages and a page giving direction, and a value indicating the area range of the document image pattern inputted from the document image pattern storage means is set as the page division information. The page division position is calculated by dividing by the number of division pages input from the input means, and the page division position and the page giving direction are output to the page unit character recognition processing means as page division information. Since pages can be added to the divided pages in accordance with the adding direction, it is suitable for editing magazines, books, etc. having different page adding directions in page units.

Further, it has a page division information input means for inputting the number of division pages, the page addition direction, the first addition page number and the total number of addition pages, and the document image pattern input from the document image pattern storage means. -Dividing the value indicating the area range of the page division by the number of divided pages input from the page division information input means to calculate the page division position,
By outputting the page division position, the page giving direction, and the giving page number as the page dividing information to the page unit character recognition processing means, the page numbers are automatically given to the divided pages in accordance with the page giving direction of the medium. Therefore, it is suitable for editing magazines, books, etc. having different page addition directions page by page.

Further, by providing the page division position correction means, even if the information medium is set in the image acquisition device with some deviation, the page division position can be corrected and page division can be performed.

[Brief description of drawings]

FIG. 1 is a functional block diagram showing a basic configuration of the present invention.

FIG. 2 is a block diagram showing the configuration of the first embodiment.

FIG. 3 is a schematic diagram of a document image pattern according to the first embodiment.

FIG. 4 is a histogram of the document image pattern shown in FIG.

FIG. 5 is a flowchart for explaining the operation of the first embodiment.

FIG. 6 is a block diagram showing a configuration of a second embodiment.

FIG. 7 is a schematic diagram of a document image pattern according to a second embodiment.

8 is a histogram of the document image pattern shown in FIG.

FIG. 9 is a flowchart for explaining the operation of the first embodiment.

[Explanation of symbols]

1 Image reading means 2 Document image pattern storage means 3 Page division information processing means 4 Page unit character recognition processing means 5 Character / graphic information storage means

─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Shizuko Kawara 1-7-12 Toranomon, Minato-ku, Tokyo Within Oki Electric Industry Co., Ltd. (72) Inventor Mitsuru Takiguchi 1-7-12 Toranomon, Minato-ku, Tokyo Oki Denki Kogyo Co., Ltd. (72) Inventor Takayuki Kato 1-7-12 Toranomon, Minato-ku, Tokyo Oki Denki Kogyo Co., Ltd. (72) Inventor Sachiyo Kurosawa 1-12-12 Toranomon, Minato-ku, Tokyo Oki Electric Industry Co., Ltd. (72) Inventor Masafumi Shimoyama 1-7-12 Toranomon, Minato-ku, Tokyo Oki Electric Industry Co., Ltd. (56) Reference JP-A-5-298477 (JP, A) JP HEI 5-135201 (JP, A) JP-A-3-201177 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) G06K 9/00-9/82

Claims (7)

(57) [Claims] 1. An information medium is optically read by an image reading means to store a document image pattern in a document image pattern storage means, and character / image separation, line segmentation, character segmentation, etc. in the document image pattern. In the optical character recognition device which performs the pre-processing and the character recognition processing and stores it in the character / graphic information storage means, a value indicating the area range of the document image pattern is input from the document image pattern storage means, and the processing is performed. The page division information processing means for calculating page division information and the document image pattern from the document image pattern storage means are inputted, and the document image pattern is subjected to pre-processing and character recognition processing to obtain the page division information. An optical character recognition device, comprising: a page unit character recognition processing means for outputting character and graphic information to the character and graphic information storage means based on page. 2. The page division information processing means inputs the document image pattern from the document image pattern storage means and calculates the number of divided pages from the shaded areas between the pages of the document image pattern,
The optical character recognition according to claim 1, further comprising: dividing a value indicating the area range of the document image pattern by the number of divided pages to calculate a page division position, and outputting the page division information to the page unit character recognition processing means. apparatus. 3. The page division information processing means has page division information input means for inputting the number of divided pages, and a page indicating a value indicating the area range of the document image pattern input from the document image pattern storage means. 2. The optical character recognition device according to claim 1, wherein a page division position is calculated by dividing by the number of divided pages input from the division information input means, and is output to the page unit character recognition processing means as page division information. 4. The page division information processing means has page division information input means for inputting the number of divided pages and a page giving direction, and the area range of the document image pattern inputted from the document image pattern storage means. The value indicating is divided by the number of divided pages input from the page division information input means to calculate the page division position, and the page division position and the page addition direction are output to the page unit character recognition processing means as page division information. Item 1. The optical character recognition device according to item 1. 5. The page division information processing means has page division information input means for inputting the number of divided pages, the page giving direction, the first added page number and the total number of added pages, and stores the document image pattern. The page division position is calculated by dividing the value indicating the area range of the document image pattern input from the means by the page division information input from the page division information input means, and the page division position and the page addition direction are calculated as the page division information. The optical character recognition device according to claim 1, wherein the assigned page number and the page unit character recognition processing means are output. 6. The optical character recognition apparatus according to claim 3, 4, or 5, wherein the divided page size is input instead of the number of divided pages. apparatus. 7. The page division information processing means divides the value indicating the area range of the document image pattern by the number of divided pages to obtain a temporary page division position, and the temporary page division position is determined from the temporary page division position. Claim 2 or Claim 3 or Claim 4 or Claim 5 or Claim 5 provided with a page division position correcting means for obtaining
Alternatively, the optical character recognition device according to claim 6.