JP2829002B2 - Character recognition device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial application field) The present invention relates to a character recognition device capable of preventing unintentional registration of a similar pattern in a dictionary and improving recognition performance for an input character pattern.

(Prior Art) With the development of information processing technology, the amount of data handled by a computer system has become enormous. Therefore, how to efficiently input various data into a computer has become a major issue.

As one of the answers to this, a character / symbol printed or handwritten on a form or the like is optically read, and the character is recognized and input as data by OCR or a handwriting input via a coordinate input device such as a tablet. Various character recognition devices have been developed which recognize characters in real time from stroke information of characters and input data.

This type of character recognition apparatus basically performs dictionary matching by calculating the similarity between an input character pattern and a dictionary pattern of various recognition target characters registered in advance in a recognition dictionary,
The recognition target character indicated by the input character pattern is obtained as a recognition result (recognition candidate) from the collation result.
However, it is not possible to sufficiently cope with deformation of a character pattern or the like only by collating various recognition target characters prepared in advance as a recognition dictionary with a dictionary pattern, and there is a problem in improving the recognition performance.

Therefore, recently, it has been considered that a personalized writing character pattern is registered in a dictionary to cope with the deformation of the character pattern, thereby improving the recognition performance. However, when character patterns unique to individuals are sequentially registered in the dictionary in this manner, similar patterns may be registered in the dictionary as dictionary patterns of different recognition target characters (different categories), and erroneous recognition may be performed between these categories. Is more likely to occur. That is, the similar pattern becomes a factor of the degradation of the recognition performance.

(Problems to be Solved by the Invention) In a character recognition device in which an input character pattern is registered as a dictionary to enhance a recognition dictionary, similar character patterns are registered as dictionary patterns of different categories. Then, there is a problem that the recognition performance is degraded by this.

The present invention has been made in view of such circumstances,
An object of the present invention is to provide a character recognition device that prevents a dictionary of similar patterns from being registered in different categories beforehand and maintains its recognition performance at a sufficiently high level.

[Structure of the Invention] (Means for solving the problem) A character recognition device according to the present invention provides information indicating that a character pattern registered in a dictionary can be registered in another category in a dictionary. A similar pattern table is prepared based on the result of matching with the dictionary pattern obtained at the time of dictionary registration, and this is stored as similar pattern information. When similarity is found from the similar pattern table, only one of the character pattern to be registered in the dictionary or the character pattern already registered in the dictionary is permitted to be registered in the dictionary. It is.

That is, when registering a new dictionary of a character pattern is prohibited, or when registering a new dictionary is permitted, a similar pattern already registered in the dictionary is registered and deleted, and the existence of a similar pattern as a dictionary pattern is eliminated. It is characterized by the following.

(Operation) According to the present invention having such a dictionary registration function, when a character pattern similar to a character pattern already registered in the dictionary is to be registered in another category in a dictionary, the similar pattern is already referred to by referring to the similar pattern table. Since it is indicated that the pattern is registered in the dictionary in another category, and only one of the patterns is permitted to be registered in the dictionary, it is possible to prevent the similar pattern from being registered in the dictionary in a different category.

As a result, erroneous recognition based on similar patterns can be prevented, and the recognition performance can be improved.

Embodiment Hereinafter, a character recognition device according to an embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a schematic configuration diagram of an embodiment apparatus, and 1 is a character pattern input unit for inputting a character pattern to be recognized (registered). The character pattern input unit 1 optically reads characters or symbols printed or handwritten on a form or the like, binarizes the read input image to obtain a character pattern, or a coordinate input device such as a tablet. Stroke information of a character input by handwriting is realized as character pattern information and a device.

The recognition processing unit 2 obtains a similarity of the character pattern input from the character pattern input unit 1 with each dictionary pattern of a plurality of recognition target characters (categories) registered in the character pattern dictionary 3, and obtains a dictionary. Characters to be recognized (category) of the dictionary pattern that obtained the highest similarity by matching
The name is obtained as a recognition result (recognition candidate) for the input character pattern. The recognition result (recognition candidate) thus determined is output from the recognition result output unit 4.

Thus, the dictionary registration unit 5 is started when the dictionary pattern registration mode is set, and inputs a character pattern given from the character pattern input unit 1 as a dictionary registration target. The dictionary registration unit 5 stores the input pattern as a dictionary pattern of a specified category.
It is to be registered. When the dictionary is registered, the dictionary registration unit 5 first refers to the similar pattern table 6 as shown in FIG. It is checked whether the dictionary is registered as a category dictionary pattern (step a). And when it is indicated that the similar pattern is already registered in the dictionary in another category, for example, the input character pattern to be registered,
The registered dictionary pattern of the dictionary category in which the similar pattern is registered is displayed (step b), and the dictionary registration of the input character pattern is stopped, or the similar pattern already registered in another category is registered in the dictionary. And prompts the dictionary registrant (operator) to select whether to newly register the input character pattern in the dictionary.

According to the information input in response to the selection instruction request, the dictionary registration unit 5 determines whether or not to newly register the input character pattern in the dictionary (step c). After a while
The registration information of the similar pattern table 6 is updated according to the dictionary registration status of the character pattern (step d).

Here, the similar pattern table 6 will be described. The similar pattern table 6 stores information based on whether or not a dictionary-registered character pattern has a possibility of being registered as a dictionary pattern of another category. It has become. Specifically, as shown in FIG. 2, assuming that a number pattern handwritten for the numbers “1”, “2”, “3”,... Is registered in a dictionary, the number pattern “2” handwritten for the number category “2” is registered. When the similar pattern table 6 is used as similar pattern information (for example, a pointer), the fact that the numeric pattern has the possibility of being registered in the dictionary for the numeric category “3” at the time of registration in the dictionary
To be stored. This indicates that there is a possibility that a character pattern similar to the handwritten number pattern "2" is registered and instructed as a dictionary pattern of the number category "3".

Note that which category of dictionary pattern is likely to be registered can be determined, for example, by performing recognition processing on a character pattern to be registered in the dictionary and examining recognition candidates. In this case, since the numeric categories "2" and "3" are obtained as recognition candidates, similar pattern information is obtained assuming that the numeric category "3" may be registered.

In a state where the numeric pattern "2" is registered in the dictionary in the numeric category "2" in this way, if the user tries to register the handwritten character pattern "3" in the numeric category "3", the similar pattern information is used. This indicates that a character pattern similar to the handwritten character pattern "3" exists in the character patterns registered in the dictionary for the number category "2". Therefore, these character patterns are presented, the operator is asked which character pattern is to be prioritized, and selection information corresponding to the character pattern is received, thereby determining whether or not to newly register the input character pattern in the dictionary. When registering a dictionary, the character pattern already registered in the dictionary is deleted.

Thus, according to the present apparatus having such a function, it is possible to prevent a similar pattern from being registered in a dictionary in a different category beforehand. As a result, erroneous recognition based on similar patterns can be prevented, and the recognition performance can be sufficiently improved.

Note that the present invention is not limited to the above-described embodiment. Here, the dictionary registration of the number pattern is described as an example, but the present invention can be similarly applied to the dictionary registration of other character types. Further, it is needless to say that the configuration of the similar pattern table 6 and the management form of the similar pattern can be variously modified. In addition, the present invention can be variously modified and implemented without departing from the gist thereof.

[Effects of the Invention] As described above, according to the present invention, it is possible to prevent a dictionary from being registered in a similar pattern in a different category, thereby improving the recognition performance of the dictionary.

[Brief description of the drawings]

FIG. 1 shows a character recognition device according to an embodiment of the present invention. FIG. 1 is a schematic configuration diagram of the device of the embodiment, FIG. 2 is a diagram showing a configuration example of a similar pattern table, and FIG. It is a figure showing the flow of processing. 1 ... Character pattern input unit, 2 ... Recognition processing unit, 3 ...
Character pattern dictionary, 4 ... Recognition result output unit, 5 ... Dictionary registration unit, 6 ... Similar pattern memory.

Claims (1)

(57) [Claims] 1. A character recognition apparatus for comparing an input character pattern with a dictionary pattern registered in a recognition dictionary in advance to obtain a recognition result for the input character pattern. A similar pattern table that stores information indicating that there is a possibility that a character pattern may be registered in another category, and a character pattern similar to the character pattern to be registered in the dictionary is already registered in another category. Character recognition characterized by comprising a dictionary registration means for permitting only one of the character pattern to be registered in the dictionary and a character pattern already registered in the dictionary when indicated from the similar pattern table. apparatus.