JPS59119486A - recognition device - Google Patents

Abstract

PURPOSE:To reduce the load of an operator and at the same time to improve a processing speed by using a similar character dictionary to recognize a pattern in case the recognition is impossible for an input pattern owing to the resemblance of features between the input pattern and the standard pattern. CONSTITUTION:The character categories are limited only to the common KANJI (Chinese character) for the reference character pattern which is stored to a standard dictionary. Thus the recognition is impossible for the shown non-common KANJI. Thus each character pattern is previously stored to a resemblance character dictionary 7 to the shown characters together with KANJI ''MIGI'' (right), etc. corresponding to ''INU'' (dog). Then ''INU'', ''MIGI'', etc. are obtained at a discriminating part 3 in order to higher degrees of resemblance to the shown characters. If these characters have the lower degree of resemblance than the prescribed value, a selection control part 5 extracts the patterns of characters similar to the shown characters stored in the dictionary 7, i.e., ''MIGI'', etc. via a retrieving part. These extracted patterns are displayed at a display part 8. An operator selects the shown characters among those displayed ones and feeds them by the key. Thus the result of recognition is obtained.

Description

DETAILED DESCRIPTION OF THE INVENTION (5) Technical Field of the Invention The present invention relates to recognition devices such as character recognition devices and back door recognition devices.

CB) Technical Background For example, in a character recognition device that recognizes a human character pattern by comparing the characteristics of an input character pattern with the characteristics of a standard character pattern stored in a dictionary, as a means to improve the recognition rate, It is conceivable to increase the number of character categories of standard character patterns stored in dictionary 4A', but such a method
This not only increases W but also reduces recognition speed.

For this reason, the character categories stored in the separate dictionary are usually limited to the characters that are used in daily life (approximately 2000 yuzu).
A method is used in which if recognition is not possible when recognition is performed using f'lJ4, other methods are used to compensate.

(C) Prior Art and Problems Conventionally, as a means to compensate for the inability to recognize characters due to restrictions on the character categories stored in the registrar, a method has been used in the past in which an operator performs all the recognition on his/her behalf and inputs the results using a character code or the like. There is. However, such a method has the drawbacks of not only increasing the burden on the operator but also not improving the processing speed.0 (C) Purpose of the Invention The purpose of the present invention is to compensate for the inability to recognize due to restrictions on the categories stored in the dictionary. As a means, it is an object of the present invention to provide a recognition device equipped with means that can reduce the burden on an operator and improve processing speed.

(G) Structure of the Invention The recognition device according to the present invention comprises: a standard dictionary that stores the features of standard patterns for each category; and means for calculating the 11 similarities between the features of the input pattern and the features of the standard pattern. In the recognition device, a similarity detail (!) is provided in which patterns of other categories similar to the standard pattern stored in the standard dictionary are stored in correspondence with the standard pattern, and the characteristics of the input pattern and the characteristics of the standard pattern are If it is not possible to identify the entry kataan due to the degree of similarity, the similarity dictionary is used to recognize the entry kataan.

[F] Embodiments of the Invention Hereinafter, embodiments of the recognition device according to the present invention will be specifically described using a character recognition device.

The figure shows a system block diagram of an embodiment of the present invention, in which 1 is printed on a sheet of paper. 2 is a feature extraction unit that extracts features of the input character pattern from the binary signal obtained by observation unit 1; 3 is a feature extraction unit that extracts features of the input character pattern from the binary signal obtained by observation unit 1; An identification unit that identifies the input character pattern by dividing the degree of similarity between the characteristics of the input character pattern and the characteristics of the standard character pattern stored in the standard dictionary described later; 4 is the standard that stores the characteristics of the standard character pattern for each character category; Dictionary, 5 is I! in the identification section 3! When the similarity history between the features of the input character pattern to be t and the features of the standard character pattern does not reach a predetermined value and the human character pattern cannot be recognized, the standard character pattern with the highest degree of similarity with the input character pattern is selected. A search section 7 searches for the corresponding similar character dictionary from other similar character categories, and 7 is the standard i! A similar character dictionary stores character patterns of other character categories similar to the standard character pattern stored in the i\book 4 in correspondence with the standard character pattern; 8 is a search unit 6'4f;
: This is a display section that displays character patterns extracted from the similar character dictionary.

If the character category of the standard character pattern stored in the standard dictionary 4 is limited to common kanji for kanji, for example, the emergency kanji ``尤J'' cannot be recognized using only 4 苧 quasi frf, calligraphy 4. It may be mistakenly recognized as 1-dog' or 'right'.

So, for example, when the corresponding expression ``尤'' is given as an input character pattern for ``inu'' for ``尤'', in 4i, printing section 3 has a high degree of similarity as a character category based on the similarity to ``尤''. Suppose that the song ni 1 dog", 1 koku", etc. are heard, but the degree of similarity in both cases is lower than a predetermined value.

In such a case, the selection system side 1 section 5 selects ``尤'' and ``Ij'' that have been previously stored in the similar character dictionary 7 in correspondence with ``dog'', which has the lowest degree of similarity to ``尤''. . . . character patterns of similar characters for temple are retrieved via the search unit 6 and displayed on the display unit 8.

The operator selects "Yu" from among the similar characters displayed on the display section 8 and presses a key to obtain the recognition result.

The display on the display section 8 - is now automatically performed.
The operator only has to select and key in the displayed similar characters. Therefore, the burden on the operator is extremely small, and the process can be completed in an extremely short time. Although this embodiment has been described with respect to a character recognition device, similar effects can be obtained even when applied to a voice recognition device.

(G) As described in detail, according to the present invention, it is possible to obtain a recognition device that reduces the burden on the operator and improves processing speed.

[Brief explanation of the drawing]

Claims (1)

[Claims] In a recognition device that includes a standard dictionary that stores features of standard patterns for each recognition category, and means for calculating the degree of similarity between the features of an input pattern and the features of the standard pattern, A similar book is provided that stores patterns of other categories similar to the standard pattern in correspondence with the standard pattern, and
A recognition device 0 whose sexual characteristic is to recognize the person's kavataan using the X word idiom, orange, when the person's kabataan cannot be recognized due to the similarity of the person's 11 characteristics with the features of the self-4 quasi-pattern.