JPS6337490A - Character recognizing device - Google Patents

Abstract

PURPOSE:To correctly recognize characters on documents including characters in similar shapes by specifying the area to be recognized of a picture to be recognized and designating the type of character in an area. CONSTITUTION:First, an input picture is binarized and stored in a picture memory part 2. A recognized area specification part 3 specifies the area to be recognized of the input picture by a rectangle shown by oblique lines. A character type specification part 4 designates the type of character in each area that the part 3 specifies as follows: A alphameric, B KANJI (Chinese character)/KANA (Japanese syllabary)/ alphameric, C-E numeric, F KANJI/KANA/alphameric, G numeric. A character segment part 5 segments picture data corresponding to the specified area one by one as a character pattern in a rectangular shape R out of the memory part 2. A feature extraction part 6 sets the directional code of a character pattern, checks the linkage of picture elements by the code, extracts strokes and further extracts a feature amount fj such as the number of strokes, position and length. A classification part 7 obtains a distance between the standard feature amount of each character in a recognition dictionary 8 and fj, takes the character with a minimum distance and the specified type for a 1st candidate character and displays said character.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to a character recognition device that recognizes printed and handwritten characters from newspapers, magazines, etc. and converts them into information amounts such as JIS codes. Conventional character recognition devices calculate the distance between the feature amount obtained from the recognition target character and the feature amount stored in advance for each character in a recognition dictionary, and select the character with the smallest distance from the recognition dictionary. were used as recognition candidate characters (for example,
IEICE General Conference 157 completed ゛Handwritten character recognition using contour direction density and background density'' Takenouchi et al.) 0
Problems to be Solved by the Invention However, in reality, in a standard document such as a form, even though only numbers are written in a certain area,
When calculating the distance between the feature amount of the recognition target character and the dictionary, calculate the distance from all characters stored in the recognition dictionary,
Since the character with the shortest distance is used as the recognition candidate character, if the character in the recognition area is like "○", it will not be possible to distinguish between the alphabet rOJ and the number rOJ, which are characters with similar shapes.
A recognition error occurred. The present invention aims to solve the above-mentioned problems, and aims to provide a character recognition device that can more accurately perform character recognition even for documents containing characters with similar shapes. .

Means for Solving the Problems In order to solve the above-mentioned problems, the present invention specifies a recognition target area of a recognition target image and specifies the character type of the recognition target area. When finding candidate characters for recognition using the distance from the feature amount of the character pattern, only characters that correspond to the character type in the recognition target area are selected as candidate characters for recognition, thereby solving the above problem.

Operation The present invention enables accurate character recognition even for similar characters such as the alphabetic character rOJ and the number "0" by the above-mentioned technical means.

Embodiments Hereinafter, embodiments of the present invention will be described with reference to the drawings. ◎ FIG. 1 is a block diagram of an embodiment of a character recognition device according to the present invention. Reference numeral 1 denotes an image input unit which scans an image including a character pattern to be recognized, inputs the image in binary code, and stores the image in the image memory unit 2. Reference numeral 3 denotes a recognition area specifying section, which specifies an area to be recognized in the input image. Reference numeral 4 denotes a character type specifying section, which specifies the character type of each recognition area specified by the recognition area specifying section 3. 5 is a character cutting part;
A rectangular character is cut out from the recognition area. Reference numeral 6 denotes a feature extracting section, which obtains feature quantities such as strokes of characters cut out by the character cutting section 5. The classification section calculates the distance between the feature amount of the recognition target character pattern obtained by the feature extraction section 6 and the feature amount of each character in the recognition dictionary 8, and selects the character type with the smallest distance among the distances. A character corresponding to the character type designated by the designation section 4 is set as a recognition candidate character. A display section 9 displays the recognition results obtained by the classification section 7.

The character recognition device configured as described above will be explained using an input image shown in FIG. 2 as an example.

An image as shown in FIG. 2 input from the image input section 1 is binarized and stored in the image memory section 2. The recognition area specifying unit 3 specifies the area to be recognized in the input image stored in the image memory unit 2 in a rectangular shape as shown in FIG. The character type specifying unit 4 specifies the character type of the recognition target area specified by the recognition area specifying unit 3, as shown in FIG. The character cutting section 5 stores the image data corresponding to the recognition target area specified by the recognition area specifying section 3 in the image memory section 2.
A rectangle R as shown in FIG. 5a is cut out one character at a time as a character pattern to be recognized.

The feature extraction unit 6 determines whether or not the character pattern P to be recognized as shown in FIG. 6 obtained by the character extraction unit is continuous in M or more including the pixel of interest in the direction indicated by the arrow in FIG. 6. is checked, a direction code is set, and strokes are extracted by checking the connectivity of each pixel for each direction code, and feature quantities 1j such as the number, position, and length of strokes are extracted. FIG. 6 shows the stroke extraction results of the character pattern rOJ.

The classification unit 7 uses the feature amount f of the character pattern to be recognized obtained by the feature extraction unit 6 and the standard feature i of each character in the recognition dictionary 8.
The distance Dk from q kj is determined by Dk=Σl f, qk51, and the character G that has the smallest value among Dk and also corresponds to the character type specified in the character type designation section 4 is found. Let n be the first candidate character.

The display unit 9 displays the candidate characters obtained by the classification unit 7.

For example, the character types in the recognition target area are C and D in FIG.

If a number is designated as E or G, and the first candidate of the recognition result for the recognition target character pattern "0" in FIG. 5 is the alphabetic character rOJ, and the second candidate is the number "0", Depending on the specified character type, the recognition result can be easily determined as the number "0".

Effects of the Invention According to the present invention, when recognizing a character pattern to be recognized, similar characters such as the alphabet rOJ and the number rOJ can be recognized with a high probability, improving the accuracy of character recognition. I can do it.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a character recognition device according to an embodiment of the present invention, FIG. 2 is an explanatory diagram showing an example of an input image, and FIG. 3 is an explanatory diagram showing a recognition target area specified for the input image. FIG. 4 is an explanatory diagram showing the designation of the character type of the recognition target area, FIG. 5 is an explanatory diagram showing the extraction of the recognition target character pattern, and FIG. 6 is an explanatory diagram showing the extraction of strokes. 1... Image input section, 2... Image memory section, 3... Recognition area specification section, 4... Character type specification section, 6...・Character cutting part, 6...
... Feature extraction section, 7... Classification section, 8...
...Recognition dictionary section, 9...Display section. Name of agent: Patent attorney Toshio Nakao and 1 other person 2nd
Figure 3 A--Inner Body Layer 8-Eleven Inner Officials/CA Rajiba/Shizukuyo Jota C---Number Sa D-Eleven Square E-Eleven Number Officials v, 4 o
p----1i3/CA et al'
y',,;/*'e3Ct---Customer Figure 5 (Ku) R Figure 6 ■

Claims (1)

[Claims] an image input section for inputting an image containing characters to be recognized; a recognition area specifying section for specifying a region to be recognized from the image input by the image input section; and a character type specifying section for specifying a character type in the recognition area. , a character cutting section that cuts out a character pattern in a region specified by the recognition area specifying section into a rectangular shape, a feature extraction section that obtains character features of the recognition target character pattern obtained by the character cutting section, and a feature amount of each character. The distance between the feature amount of the character pattern to be recognized obtained by the feature extraction section and the feature amount of each character stored in the recognition dictionary in advance is determined, 1. A character recognition device comprising: a classification section that sets a character with the smallest distance among the character types specified by the character type specification section as a recognition candidate character.