JPH0773274A - Character recognition device - Google Patents

Abstract

PURPOSE:To provide a character recognition device which excels in the reliability, the workability and the universal applicability and can easily and correctly recognize a relevant document even if the character directions are not fixed or inverted and also can recognize even such characters that are set in different directions or inverted. CONSTITUTION:A character recognition device is provided with an image reading part 1 which acquires the binary image data, a character image segmenting part 3 which segments the character image data out of the image data, a character image data storing part 4, a character image data left 90 deg.C turning part 5, a character feature dictionary 6 which stores the normal features of all characters, a recognizing precision calculating part 7 which refers to the dictionary 6 to aclculate the precision of the candidate characters resemble to the character image data and their resemblance, etc., and a recognizing character deciding part 8 which decides the recognizing characters based on the recognizing precision.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character recognition device for converting printed characters, dot characters and handwritten character patterns of newspapers, magazines, novels etc. into code information such as JIS code.

[0002]

2. Description of the Related Art In recent years, with the development of office automation, a character recognition device is used to input a printed manuscript or the like to a computer or the like without using a keyboard.

A conventional character recognition device will be described below. FIG. 8 is a diagram showing an example of a recognition target document, and FIG. 9 is a diagram showing an example of a recognition target document whose front and back are reversed.

In the conventional character recognition apparatus, when the document to be recognized is as shown in FIG. 8, the character of part a can be recognized as it is because its orientation matches the orientation of character recognition. Although it is possible, the direction of the character of part b is rotated 180 degrees from the direction of character recognition, and the direction of the character of part c is also rotated 90 degrees to the right.
I can't recognize. Therefore, first, the character of the portion a is recognized, and then the user rotates the recognition target document so that the direction of the character of the portion b matches the direction of character recognition, and the character of the portion b is recognized. Recognizing, and then recognizing the character c in the same manner, and then the user integrates these recognition results to recognize the entire recognition target document.

Further, when recognizing a document written on a transparent sheet such as an OHP sheet, when the sheet is mistakenly set upside down, the characters as shown in FIG. 9 are reversed. Documents that consist of are the recognition target documents. In this case, since the front and back of each character are reversed, character recognition cannot be performed at all. Therefore, in order to perform character recognition, it is necessary for the user to correctly set the front and back sides of the sheet.

[0006]

However, in the above-mentioned conventional configuration, when the orientation of the characters in the document to be recognized is not constant, recognition is performed for each orientation of the characters and the user integrates the recognition results. Therefore, there is a problem in that it is complicated and time-consuming, lacks workability and versatility, and may cause misrecognition, and lacks reliability. In addition, when the document to be recognized is set upside down, it is not possible to recognize characters that are upside down, resulting in a lack of workability and versatility, and the possibility of erroneous recognition resulting in a lack of reliability. Had.

The present invention solves the above-mentioned problems of the prior art, and the orientation of the characters in the document to be recognized is not constant,
Even if the front and back of the characters are reversed, the document to be recognized can be correctly recognized and it is highly reliable, and the recognition work can be performed easily and the workability is excellent. It is an object of the present invention to provide a versatile character recognition device capable of recognizing even characters.

[0008]

In order to achieve this object, a character recognition apparatus according to claim 1 of the present invention comprises an image reading unit for converting a document to be recognized into binary image data, and the image reading unit. Character recognition including a character image cutout unit that cuts out character image data of one character from the binary image data output from the unit, and a character image storage unit that stores the character image data cut out by the character image cutout unit A device, a character image conversion unit for rotating and / or reversing the character image data stored in the character image storage unit, a character feature dictionary storing normal character features of all characters, and the character A recognition accuracy calculation unit that obtains a recognition accuracy of a candidate character similar to the character image data stored in the character image storage unit and the similarity by referring to the feature dictionary, and a recognition accuracy calculation unit that calculates the recognition accuracy. A recognition character determining unit that determines the recognized character from the recognition certainty which is has a configuration including a
The character recognition device according to claim 2 is an image reading unit that converts a recognition target document into binary image data, and a character that cuts out one character image data from the binary image data output from the image reading unit. A character recognition device comprising: an image clipping unit; and a character image storage unit for storing the character image data cut out by the character image clipping unit, wherein the character feature dictionary stores normal character features of all characters. And a converted character feature dictionary storing converted character features obtained by rotating and / or reversing the normal character features of all characters,
A recognition accuracy calculation unit that obtains recognition accuracy of candidate characters similar to the character image data stored in the character image storage unit and the similarity, by referring to the character feature dictionary and the converted character feature dictionary; And a recognition character determination unit that determines a recognition character from the recognition accuracy obtained by the recognition accuracy calculation unit.

[0009]

With this configuration, the character image data stored in the character image storage unit is rotated / rotated by the character image conversion unit.
While reversing the front and back, the recognition accuracy calculation unit refers to the character feature dictionary to obtain the recognition accuracy, so that the direction of the character is not constant or the front and back of the character are reversed without increasing the memory required for the character feature dictionary. It is possible to correctly and easily recognize the recognition target document. Also,
By installing a converted character feature dictionary that rotates the normal character features of all characters and reverses the front and back, character image data in which the direction of the character is not constant or the front and back of the character are reversed , And can be recognized correctly, easily and at high speed without conversion by the character image conversion unit or the like.

[0010]

【Example】

(Embodiment 1) A character recognition apparatus according to an embodiment of the present invention will be described below with reference to the drawings. 1 is a functional block diagram of a character recognition device according to an embodiment of the present invention, and FIG. 2 is a circuit block diagram of the character recognition device according to an embodiment of the present invention. In FIG. 1, reference numeral 1 denotes an image reading unit configured by a scanner (not shown) or the like for reading a recognition target document and outputting image data, and 2 denotes image data for one page stored from the image data output from the image reading unit 1. A page image storage unit 3, a character image cutout unit that cuts out character image data for one character from one page of image data stored in the page image storage unit 2, and 4 a character cut out by the character image cutout unit 3. A character image storage unit 5 for storing image data and the like is an example of a character image conversion unit, and the character image data stored in the character image storage unit 4 is moved to the left 9
A character image rotated 90 degrees to the left by 90 degrees, 6 is a character feature dictionary that stores normal character features of all characters, and 7 is stored in the character image storage unit 4 with reference to the character feature dictionary 6. Recognition accuracy calculation unit for obtaining recognition accuracy of candidate characters similar to the character image data and similarity, and 8 is a recognition character determination unit for determining the recognition character and the rotation angle from the recognition accuracy output from the recognition accuracy calculation unit 7. Is. In FIG.
Reference numeral 9 is a CPU that controls the entire character recognition device, 10 is a scanner that constitutes the image reading unit 1 and converts a recognition target document into image data, and 11 is a page image storage area 12 that serves as a page image storage unit 2 and a character image storage unit. A RAM having a character image storage area 13 serving as the unit 4 has a character feature information storage area 15 for storing the character feature dictionary 6 and the like, and a program storage area 16 for storing a program for instructing all processes in the CPU 9. ROM.

The operation of the character recognition device having the above-described structure according to the embodiment of the present invention will be described below. FIG. 3 is a flowchart of the character recognition device in one embodiment of the present invention, and FIG. 4 is a diagram showing an example of recognition processing by the character recognition device in one embodiment of the present invention.

First, the image reading section 1 reads a document to be recognized, converts it into image data, and stores the image data for one page in the page image storage section 2 (S).
1). Next, the character image cutout unit 3 cuts out character image data for one character from the image data stored in the page image storage unit 2 and stores it as original character image data in the character image storage unit 4 (S2). Next, the recognition accuracy calculation unit 7 refers to the character feature dictionary 6 to refer to the candidate character for the original character image data in the character image storage unit 4,
The recognition accuracy such as the similarity is calculated as usual (S3). Next, the character image left 90 ° rotation unit 5 rotates the character image data in the character image storage unit 4 90 ° to the left (S
4). Next, the recognition accuracy calculation unit 7 refers to the character feature dictionary 6 and extracts candidate characters, similarity degrees, etc. for the character image data obtained by rotating the original character image data 90 degrees to the left in the character image storage unit 4. The recognition accuracy is obtained (S5). Next, the character image left 90-degree rotation unit 5 rotates the character image data in the character image storage unit 4 further 90 degrees to the left (S6). Next, the recognition accuracy calculation unit 7 refers to the character feature dictionary 6 and recognizes the recognition accuracy of the candidate character, the similarity, and the like for the character image data obtained by rotating the original character image data 180 degrees in the character image storage unit 4. Is calculated (S7). Next, the character image left 90-degree rotation unit 5 rotates the character image data in the character image storage unit 4 further 90 degrees to the left (S8). Next, the recognition accuracy calculation unit 7 refers to the character feature dictionary 6, and refers to the character image data in the character image storage unit 4 by rotating the original character image data to the left by 270 degrees, such as candidate characters and similarity. Obtain recognition accuracy (S
9). Next, the recognized character determination unit 8 causes S3, S5,
Among the candidate characters obtained in S7 and S9, the character having the highest degree of similarity is set as the recognition result, and at the same time, its rotation angle is also determined (S10). Next, it is checked whether all the characters stored in the page image storage unit 2 have been recognized (S11). If No, jump to S2 to recognize the next character,
If Yes, it is checked whether all pages of the recognition target document have been recognized (S12). If No, jump to S1 in order to recognize the next page, and if Yes, end all processing.

Here, as an example, when the original character image data is as shown in FIG. 4, the character recognition device in one embodiment of the present invention rotates the image 90 degrees to the left and recognizes the recognition accuracy. As is clear from FIG. 4, the similarity to the candidate character'F 'when the original character image data is rotated 270 degrees to the left is the highest. Therefore, it is determined that the original character image is obtained by rotating ‘F’ to the right 270 degrees, and this is set as the recognized character.

In the present embodiment, the character image conversion unit is the character image left 90-degree rotation unit 5 which rotates 90 degrees to the left, but this may be another direction or angle. May be reversed, or both may be performed simultaneously.

As described above, according to this embodiment, by providing the character image conversion unit, the character is rotated or the front and back of the character are reversed without increasing the memory required for the character feature dictionary 6. It is possible to correctly recognize the characters that are moved.

(Second Embodiment) A character recognition apparatus according to a second embodiment of the present invention will be described below with reference to the drawings. FIG. 5 is a functional block diagram of the character recognition device in the second embodiment of the present invention. 1 is an image reading unit, 2 is a page image storage unit, 3 is a character image cutout unit, 4 is a character image storage unit, 6 is a character feature dictionary, 8 is a recognized character determination unit, and these are the same as those in the first embodiment. Therefore, the same reference numerals are given and the description thereof will be omitted. Reference numeral 17 is an example of a converted character feature dictionary, which is a left 90-degree rotated character feature dictionary that stores left 90-degree rotated character features that are rotated by 90 degrees to the left for normal character features of all characters, and 18 is left 90-degree. A 180-degree rotated character feature dictionary that stores 180-degree rotated character features similar to the rotated character feature dictionary 17, and 19 indicates a left 270-degree rotated character feature that stores left 90-degree rotated character feature dictionary 17 and a left 270 degree that stores rotated character features The rotated character feature dictionary, 20 is stored in the character image storage unit 4 with reference to the character feature dictionary 6, the left 90-degree rotated character feature dictionary 17, the 180-degree rotated character feature dictionary 18, and the left 270-degree rotated character feature dictionary 19. Is a recognition accuracy calculation unit that obtains recognition accuracy of a candidate character similar to the existing character image data and similarity.

The operation of the character recognition apparatus according to the second embodiment of the present invention having the above-described structure will be described below. FIG. 6 is a flowchart of the character recognition device in the second embodiment of the present invention, and FIG. 7 is a diagram showing an example of recognition processing by the character recognition device in the second embodiment of the present invention.

First, the image reading unit 1 reads a document to be recognized, converts it into image data, and stores the image data for one page in the page image storage unit 2 (S1).
3). Next, the character image cutout unit 3 cuts out character image data for one character from the image data stored in the page image storage unit 2 and stores it in the character image storage unit 4 (S14). Next, the recognition accuracy calculation unit 20 causes the character feature dictionary 6, the left 90-degree rotated character feature dictionary 17, 180.
While referring to the degree-rotated character feature dictionary 18 and the left 270-degree rotated character feature dictionary 19, the recognition accuracy of the candidate characters, the similarity, etc. for the character image data in the character image storage unit 4 is obtained, and the recognized character determination unit 8 determines Among these candidate characters, the character having the highest degree of similarity is set as the recognition result, and at the same time, its rotation angle is also determined (S15). Next, it is checked whether all the characters stored in the page image storage unit 2 have been recognized (S16).
If No, jump to S14 to recognize the next character, and if Yes, check whether all pages of the document to be recognized have been recognized (S17). If No, jump to S13 to recognize the next page.
If Yes, all the processes are finished.

Here, as an example, when the original character image data is as shown in FIG. 7, the character recognizing device in the second embodiment of the present invention, as shown in FIG.
Assume that the similarity to the candidate character “F” in the 0-degree rotated character feature dictionary 19 is the highest, and the original character image has “F” to the right 27
It is determined that the character has been rotated by 0 degrees, and this is set as the recognition character.

In the present embodiment, as the converted character feature dictionary, a left 90-degree rotated character feature dictionary 17, a 180-degree rotated character feature dictionary 18, and a left 270 are obtained by rotating normal character features 90 degrees to the left. Although the degree-rotated character feature dictionary 19 is used, these may be in other directions and angles, the characters may be reversed, or both may be performed at the same time.

As described above, according to the present embodiment, by providing the converted character feature dictionary, the character image data is converted into the character image data for the character that is rotated or the character is reversed. It is possible to recognize correctly, easily and at high speed without conversion by the conversion unit or the like.

[0022]

As described above, the present invention refers to the character feature dictionary by the recognition accuracy calculation unit while rotating and reversing the character image data stored in the character image storage unit by the character image conversion unit. By determining the recognition accuracy, it is possible to correctly and easily recognize a recognition target document in which the orientation of the characters is not constant or the front and back of the characters are reversed without increasing the storage capacity required for the character feature dictionary. It has excellent character, versatility, reliability, high memory efficiency, and has a converted character feature dictionary that rotates the normal character features of all characters and reverses the front and back, so that the orientation of characters is constant. There is no or the front and back sides of the characters are reversed, and the workability and general ability to recognize the character image data of the recognition target document correctly and easily and at high speed without converting it with the character image converter etc. Sex, in which an excellent character recognition device reliability can be realized.

[Brief description of drawings]

FIG. 1 is a functional block diagram of a character recognition device according to an embodiment of the present invention.

FIG. 2 is a circuit block diagram of a character recognition device according to an embodiment of the present invention.

FIG. 3 is a flowchart of a character recognition device according to an embodiment of the present invention.

FIG. 4 is a diagram showing an example of recognition processing by a character recognition device according to an embodiment of the present invention.

FIG. 5 is a functional block diagram of a character recognition device according to a second embodiment of the present invention.

FIG. 6 is a flowchart of a character recognition device according to a second embodiment of the present invention.

FIG. 7 is a diagram showing an example of recognition processing by a character recognition device according to a second embodiment of the present invention.

FIG. 8 is a diagram showing an example of a recognition target document.

FIG. 9 is a diagram showing an example of a recognition target document whose front and back are reversed.

[Explanation of symbols]

 1 image reading unit 2 page image storage unit 3 character image cutout unit 4 character image storage unit 5 character image left 90 degree rotation unit 6 character feature dictionary 7 recognition accuracy calculation unit 8 recognition character determination unit 9 CPU 10 scanner 11 RAM 12 page image Storage area 13 Character image storage area 14 ROM 15 Character feature information storage area 16 Program storage area 17 Left 90-degree rotated character feature dictionary 18 180-degree rotated character feature dictionary 19 Left 270-degree rotated character feature dictionary 20 Recognition accuracy calculation unit

Claims (2)

[Claims] 1. An image reading unit for converting a document to be recognized into binary image data, a character image cutout unit for cutting out character image data of one character from the binary image data output from the image reading unit, and the character. A character recognition device comprising: a character image storage unit that stores the character image data cut out by the image cutout unit, wherein the character image data stored in the character image storage unit is rotated and / or inverted. A character image conversion unit, a character feature dictionary that stores normal character features of all characters, a candidate character that is similar to the character image data stored in the character image storage unit by referring to the character feature dictionary, and similar characters Characters including a recognition accuracy calculation unit for determining recognition accuracy such as degree, and a recognition character determination unit for determining recognition characters from the recognition accuracy calculated by the recognition accuracy calculation unit. Identification equipment. 2. An image reading unit for converting a document to be recognized into binary image data, a character image cutting unit for cutting character image data of one character from the binary image data output from the image reading unit, and the character. A character recognition device comprising a character image storage unit for storing character image data cut out by an image cutout unit, a character feature dictionary storing normal character features of all characters, and a normal character of all characters. Stored in the character image storage unit with reference to the converted character feature dictionary and the converted character feature dictionary, which stores the converted character feature obtained by rotating and / or inverting the character feature Recognition accuracy calculation unit for obtaining recognition accuracy of candidate characters similar to the character image data and similarity, and a recognition character determination unit for determining recognition character from the recognition accuracy obtained by the recognition accuracy calculation unit. Character recognition apparatus characterized by comprising a.