JPS6010393A - Character recognizing device - Google Patents

Abstract

PURPOSE:To obtain easily a capital letter string and a small letter string by extracting information giving a change in size at an input section and giving said information after the recognition of character. CONSTITUTION:When a manually written character is inputted to an input section 1 and patternized, the character information is inputted also to a size discriminating circuit 5 and the information is divided at least into two; English capital letter and small letter: large Katakana (Japanese syllabary) and small Katakana. In this case, even when the input section intends to attain English small latter output, the section forms large letters with a low letter height and when the section intends to apply large letter output, the section forms large letters with a high height. On the other hand, the input section 1 normalizes input characters having different height, a characteristic extracting section 2 extracts the characteristic, and a dictionary 3 is utilized so as to recognize characters by a comparison recognizing section 4. When, e.g., a control signal of a small letter exists, a character output device 6 outputs small letters by using the information from the discriminating circuit 5.

Description

DETAILED DESCRIPTION OF THE INVENTION (1) Technical Field of the Invention The present invention relates to a character recognition device that can relatively easily distinguish recognition result outputs into large characters, small characters, uppercase letters, and lowercase letters.

(2) Prior art and problems Research into inputting handwritten characters and outputting recognition results is progressing. Currently, uppercase English letters, lowercase English letters, katakana characters, and Arabic numerals can be recognized with almost no errors, but these devices are becoming extremely large, meaning that they cannot recognize the desired characters as output patterns or the characters specified by the dictionary and output printing device. and enter the corresponding format characters. In Fig. 1, the input section (11) is a pattern in which the input character is a binary code string, and the feature extraction section (2) extracts the features. , the features are compared in the comparative recognition unit (4).The one with the best consistency is output as the recognition result and the same character as stored in the dictionary.For example, when the uppercase English letter A is input, only A is output, and the lowercase English letter a When inputting , the output was also a, so in order to distinguish and recognize the two, dictionary (3) needed a huge capacity.Also, since the size of the characters to be input was standardized, a device for that purpose was also required. Then, the dictionary (3) and the comparison recognition section (
4) The equipment becomes large-scale.

(3) Object of the Invention An object of the present invention is to improve the above-mentioned drawbacks and to provide a character recognition device that can relatively easily distinguish and output characters.

(4) Structure of the Invention The structure of the present invention to achieve the above-mentioned object is to pattern handwritten input characters, extract their features, compare the features with a pre-stored dictionary, and perform predetermined recognition. In a character recognition device that obtains results, the present invention controls a circuit that determines the vertical or horizontal size of handwritten input characters into a plurality of categories, and a device that distinguishes output characters of recognition results based on the output of the circuit.

(5) Embodiment of the Invention FIG. 2 shows the configuration of an embodiment of the invention, and the same reference numerals as in FIG. 1 indicate the same components. (5) is a circuit that determines the vertical size of handwritten input characters into multiple categories; (6)
indicates a character output device. Input section (as handwritten characters)
1), and when it is patterned, it is applied to the size determining circuit (5) and is divided into at least two types, such as uppercase letters, lowercase letters, and large and small letters in katakana. At this time, the input section is made of alphabetic characters, and when outputting lowercase letters is intended, the letters are uppercase and have a lower height, and when outputting uppercase letters are intended, the letters are uppercase and taller. Similarly, katakana also has tall characters.
Use low font. For comparison, in FIG. 3, the left side shows the input section, and the right side shows the corresponding output characters. The input unit (1) normalizes the input characters having different heights, and then the feature extraction unit (2) extracts the features. At this time, no distinction is made between uppercase and lowercase letters. dictionary(
The operation of the comparison recognition unit (4) using 3) is the same as the conventional method, and a recognition result can be obtained. Since the output from the size classification circuit (5) is applied to the character output device (6), for example, if there is a control signal for lowercase English letters, lowercase letters can be output. When the capabilities of the input section (1), feature extraction section (2), and dictionary (3) improve, it becomes possible to recognize hiragana and kanji. In addition to dividing by height difference, it is also possible to make the characters double or half the size in the horizontal direction. All you have to do is add a classification symbol to the input characters and have them recognized.

(6) Effects of the Invention In this way, according to the present invention, uppercase and lowercase English character strings can be easily obtained by simply extracting information that gives a change in size in the input section and controlling the character output after recognition. It will be done. For this reason, it is difficult to read uppercase English character strings, but according to the present invention, easy-to-read output can be obtained with characters mixed with lowercase characters. Since the characters prepared in the dictionary can be basic characters, the capacity is small, and since there are few contrasts to be recognized, recognition results can be obtained quickly and misrecognitions can be reduced.

[Brief explanation of the drawing]

Figure 1 is an explanatory diagram of a conventional character recognition device, Figure 2 is a diagram showing the configuration of an embodiment of the present invention, and Figure 3 is an example of the correspondence between input characters and recognized output characters compared to Figure 2. FIG. (ll----Manpower Department (2)・-Feature Extraction Unit+31-M-Dictionary (41--Comparison Recognition Unit (5)-Size Sorting Circuit (61--Single Character Output Device Patent Applicant Fujitsu Limited Agent) People Patent Attorney Eisuke Suzuki Figure 1 A-A -G -B -b Artearte Figure 3

Claims (1)

[Claims] In a character recognition device that patterns handwritten input characters, extracts their features, compares the features with a pre-stored dictionary, and obtains a predetermined recognition result, the size of the handwritten input characters in the vertical or horizontal direction is determined. What is claimed is: 1. A character recognition device that controls a circuit that determines a character into a plurality of categories, and a device that distinguishes output characters of recognition results based on the output of the circuit.