JP2570815B2 - Optical character recognition device - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to an optical character recognition device, and particularly to a clockwise
The present invention relates to an optical character recognition device capable of recognizing a form input by being rotated by 90 degrees without requiring special firmware.

[Conventional technology]

Conventionally, this type of optical character recognition apparatus inputs a form image rotated by 90 degrees with respect to the transport direction, and then performs recognition processing. Two image input methods are used as shown in FIGS. 4 (a) and 4 (b).

First, in the first method, as shown in FIG. 4 (a), the form image 61 is directly taken into a page buffer, and when the form image 61 is transferred from the page buffer to the line buffer, a counter-clockwise direction is applied to each line requiring character recognition. The image is rotated by 90 degrees and the obtained image is stored in a line buffer, and character recognition processing is performed for each line or for a plurality of lines.

The second method is to rotate the form image 80A of the page buffer 90A in the counterclockwise direction by one mesh in order of meshes of the image obtained by the scanner as shown in FIG. Take in the page buffer rotated 90 degrees. Such an image is stored in a line buffer, and character recognition processing is performed for each line or for each of a plurality of lines.

[Problems to be solved by the invention]

In the above-described conventional optical character recognition apparatus, in the first method of inputting a form image which is rotated 90 degrees clockwise and required in the above-described method, the entire form image is completely input to the page buffer in the first method. In the second method, sample data of the image data from the scanner unit is used for the page buffer. However, there is a drawback in that the time taken to capture the image into the page buffer becomes longer due to the restriction on the memory access time.

[Means for solving the problem]

In order to solve the above problems, an optical character recognition device according to the present invention comprises a scanner for scanning a form and inputting image data, and a first and a second scanner for writing image data alternately by a predetermined line from the scanner. And the first or second line buffer
Means for reading out the image data already written therefrom from the line buffer on which the image data writing operation has not been performed and rotating it by 90 degrees; image data from the scanner or the first or second image data; A page buffer for storing one page of image data read from the line buffer, and means for recognizing a pattern on a form based on the image data stored in the page buffer.

〔Example〕

 Next, the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of one embodiment of the present invention.

Referring to FIG. 1, an embodiment of the present invention includes a scanner 1
, Scan control unit 2, preprocessing unit 3, character recognition unit 4
It is composed of

When the form image signal 101 from the scanner unit 1 is not rotating clockwise, the switching unit SW1 in the scan control unit 2 is connected to the contact C, the switching unit SW2 is connected to the contact F, and the image signal 101 is Direct serial / parallel converter 2
Sent to 4. The serial / parallel converter 24 stores the image signal 104 for 8 meshes and sends it to the page buffer 26 as a signal 105 in order to improve the processing speed. In the page buffer 26, the page buffer address control unit 25 sequentially outputs the signal 105 from the head address of the page buffer 26.
To store.

Thereafter, the character pattern 106 is called from the page buffer 26, and the character pattern 107 processed by the general pre-processing unit 3 for extracting a character and normalizing or thinning the character is sent to the character recognizing unit 4 for recognition.

Next, when the form image 101 from the scanner unit 1 is output rotated 90 degrees clockwise, the switching output stage SW1 in the scan control unit 2 is first connected to the contact A, and the image signal 101 is set to A. The data is sent to the line buffer 22 and stored for eight lines. If eight lines have been stored in the A line buffer 22, the switching means SW1 is connected to the contact B, and the switching means SW2 is connected to the contact D. Thereafter, the image signal 102 is read from the A line buffer 22 by the address signal 107 controlled by the line buffer address control unit 21. The signal 105 stored for 8 meshes is sequentially stored in the page buffer 26 by the serial-parallel conversion unit 24 in accordance with the address signal 109 controlled by the page buffer address control unit 25.

During the above processing, the image signal 101 is sent to the B line buffer 23 and stored for eight lines. When the reading of the A line buffer 22 and the writing of the B line buffer 23 are completed, the switching means SW1 is again connected to the contact A, the switching means SW2 is connected to the contact E, and the image signal 101 is supplied to the A line buffer 22. Are stored in the B-line buffer 23 by the page buffer address control unit 21, the image signal 103 is read by the address signal 108 controlled by the page buffer
Sent to 4. Here, the signal 105 stored for 8 meshes is sequentially stored in the page buffer 26 by the address signal 109 controlled by the page buffer address control unit 25.

The combination in which the switching means SW1 and SW2 are connected to the contacts A and E respectively, and the combination in which the switching means SW1 and SW2 are connected to the contacts B and
The combination connected to D continues to be switched alternately until all the form images have been stored in the page buffer 26.

Next, referring to FIGS. 2 (a) and (b) and FIGS. 3 (a) and (b), the specific operation of the scan control unit 2 in FIG. The case where one address = 8 meshes are stored in the page buffer and eight lines are stored in the line buffer will be described.

When the image signal 101 from the scanner unit 1 is at the contact A of the switching means SW1, as shown by reference numeral 10 in FIG.
A line buffer address control unit in the A line buffer 22
The upper left mesh (0
000) as the origin, from left to right (0000 to 0FFF), rightmost (0
FFF), it goes down one line downward, the left end mesh is the origin, it goes down from left to right (1000-1FFF), and it goes down sequentially to the lower right mesh (7FFF) and is stored for 8 lines. . The mesh numbers in the parentheses described above are displayed in hexadecimal notation, and unless otherwise specified, are displayed in hexadecimal notation.

Next, when the contact is switched to the contact D of the switching means SW2, the upper right (0000) of the A line buffer 22 is changed according to the address signal 107 of the line buffer address controller 21 as shown by reference numeral 20 in FIG. From the top to the bottom (0000
~ 0007, ..., 7FF8 ~ 7FFF) while going from right to left (0000
~ 0008 ~ ... ~ 7FFF8) and the image signal 102 is read out. At the same time as the above reading, FIG.
Image signal 101 according to the mesh indicated by reference numeral 10
Is connected to the B line buffer 23 through the contact B of the switching means SW1.
Is read in.

The read image signal 102 is converted into 8 meshes (0000 to 0007, 0008 to 000F,.
…, 7FF8 ~ 7FFF) Then, as shown at 30 in FIG. 2 (b), the address 10 by the page buffer address control unit 25 is set with the upper left (0000) of the page buffer 26 as the origin.
9, from top to bottom (0000 ~ 0FFF, 1000 ~ 1FFF, ..., n000 ~
Going from nFFF, ...) from left to right (0000 ~ 1000 ~ ... ~ n00)
0 to...), And the image data 105 for eight meshes are sequentially stored in the page buffer 26.

If the above operation is shown more specifically, first,
Reference numeral 40 in FIG. 7A shows a form image of the scanner unit 1 rotated clockwise by 90 degrees. Also, the third
Reference numeral 50 in FIG. 3B shows a form image on the page buffer 26 which is obtained by rotating the form image of reference numeral 40 in FIG. 3A by 90 degrees counterclockwise. .. Of reference numeral 40 in FIG. 3A correspond to the reference numerals 50 in FIG. 3B, respectively.

As described above, without impairing the processing speed,
The form image that is input rotated 90 degrees clockwise,
By rotating it 90 degrees counterclockwise, a normal form image can be captured in the page buffer. In the case of reading a form rotated by 90 degrees, the form is likely to be skewed or meandered due to the transport performance, but in the present invention, the image of the entire form is stored in the page buffer. Therefore, it is possible to easily correct the image data. Therefore, since the pattern recognition is performed after the correction, the reduction of the recognition rate can be prevented.

〔The invention's effect〕

As described above, according to the present invention, the image signal is stored in the page buffer by n meshes and stored in the page buffer between the form image and the page buffer. There is an effect that ordinary character recognition processing can be performed without rotating the form image sent by 90 degrees counterclockwise without deteriorating the processing speed and storing it in the page buffer and developing dedicated firmware.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of an embodiment of the present invention, and FIGS. 2 (a) and 2 (b) are explanatory diagrams showing a line buffer address control section and a page buffer address control section of FIG. FIGS. 3 (a) and 3 (b) are explanatory diagrams showing a form image sent from a scanner unit and a result stored in a page buffer, and FIGS. 4 (a) and (b) are form images and pages according to the prior art. Explanatory drawing showing an example of a buffer storage result. 1. Scanner unit 2. Scan control unit 3. Preprocessing unit 4. Character recognition unit.

Claims (1)

(57) [Claims] A scanner for inputting image data by scanning a form; first and second run-in buffers in which image data is alternately written by a predetermined line from the scanner; and a first or second run-in buffer. Means for reading out the image data already written therein from the line buffer in which the image data write operation is not performed among the line buffers and rotating the image data by 90 degrees counterclockwise; and image data from the scanner. A page buffer for storing one page of the image eater read from the first or second line buffer, and a unit for recognizing a pattern on a form based on the image data stored in the page buffer. An optical character recognition device, comprising: