JPS63282896A - On-line hand-written character recognizing device - Google Patents

Abstract

PURPOSE:To reduce processing time by retrieving a dictionary about all character sorts at real time every input of one stroke by means of dictionary retrieving pointers, storing character codes in result storing buffers, and after detecting the end of one character input, determining a character code to be outputted. CONSTITUTION:A recognizing part has the dictionary retrieving pointers 14a-14c corresponding to the number of character sorts such as alphabet, numerals, 'KANJI' (Chinese character), and 'HIRAGANA' (cursive form of Japanese syllabary) to be recognized and the result storing buffers 10a-10e, and every input of one stroke, retrieves the dictionary about all the character sorts at real time by means of the pointers 14a-14c. A character code about the character sort obtaining the retrieved result is stored in the buffers 10a-10e, and after ending the input of one character, a character code to be finally recognized is determined by a character sort determining means 11 for specifying the character sort to be outputted. Consequently, the specification of a character sort prior to the input of a character can be omitted and the interruption of consideration due to the change of a character sort is not generated.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a recognition method in an on-line old-fashioned character recognition device.

[Conventional technology]

Among conventional character recognition devices, there is no one that can obtain a recognition rate that is sufficient for fA without specifying the type of character to be recognized. Therefore, it is necessary to specify the character type in some way, and there are two main methods: one is to specify it before entering the characters, and the other is to specify it after the characters are entered.

What is specified before writing is generally a mode specification, and once the tft is specified, the same character type will be recognized until the next specification.

If you specify a character after writing, specify the character type in the single character input mode, and recognition will be performed accordingly.

[Problem that the invention seeks to solve]

However, the above-mentioned conventional technology has the following problems.

(1) When specifying before entry, if the character type changes while inputting the sentence jy, the character type must be specified, which is extremely inconvenient.

(2) If you write without changing the character type designation, you must cancel the characters you have written.

(3) After entering the characters, specify the character type (if specified, the character type must be set to 1, then the dictionary search will start, so the character type must be set to 2).
It is likely that the time will increase. There is a way to improve the performance of the hardware to alleviate this problem, but
This poses a problem in that the product cost increases.

The present invention is intended to solve these problems, and its purpose is to provide an online handwritten character recognition device that is inexpensive and easy to use.

[Means for solving problems]

The Onoline handwritten character recognition device of the present invention includes an input unit that detects the two-dimensional coordinates of a writing point and identifies the shape of the stroke determined by on and off of the brush when inputting old-fashioned characters; A character dictionary configured hierarchically by a combination of character codes, standard II strokes constituting the characters, and segments defined by a plurality of stroke sequences, the input strokes, and the character dictionary. In a system that includes at least a recognition section that recognizes characters input by the user, the recognition section a is used for dictionary searches as many as the number of character types to be recognized, such as English letters, numbers, kanji, and hiragana. The dictionary is searched in real time for all character types each time one stroke is input, according to the dictionary search function, and the character type for which the result is obtained. , the character code is stored in a result storage buffer, and after one character has been input, the finally recognized character code is determined by a character type determining means that specifies the character type to be output.

(Embodiment) The following is an embodiment of the present invention. Fig. 1 is a diagram of the software according to the present invention. The result is stored in the stroke/pulling data buffer 7G.Subsequently, in the stroke recognition unit 3, the sampling buffer 6
Converts the coordinate data into the stroke code required for dictionary search. Here, the stroke dictionary 8 prepared in advance is referred to. The stroke hood obtained by the stroke recognition section 3 is stored in a stroke buff 77. The dictionary search section 4 searches the character dictionary based on the data in the stroke buffer. The above processing is performed every time a stroke is input. Therefore, before one character is inputted, the dictionary search has already been completed to some extent in the search section 4.

The end of character input is notified by inputting the character type designation key set to H:1. Due to this, the electric strike
- Kubasov 7 stores a special code that indicates the end of character input. Same u1? Character type specification buffer 11
stores a code that specifies the input character type.

When the dictionary search unit 4 detects the character input end code, it continues processing until the dictionary search is completed and transfers the results for each character type from the kanji buffer IOA to the English character buffer 7101.
Store by E. If there is a character type of 8 yaki fu 11 shi, a special code indicating that it is unrecognizable is stored in the character buffer. After that, the output unit 5 checks the character type (the value of the i constant buffer 11, and
The character code contained in the specified character buffer from the IOE is sent to the output buffer sofa 13.

FIG. 2 is a diagram that does not show the configuration of the 7.7'a snare section 4.

Inside the search section, there are a pointer 14a for searching hiragana, a pointer 14c for searching katakana, a pointer 14d for searching numbers, and a pointer 14d for searching English characters. 14c
is set, and each pointer indicates the search position in the character dictionary ().

Every time the data indicated by the pointer 15 indicating stroke data 1 is stored in the buffer 716, the search is performed by comparison with the dictionary data indicated by each character type pointer.

FIG. 3 is a flowchart showing the processing procedure in the dictionary search section 4.

30 Before starting dictionary search, search for each character type -1
:Reset Inota to point to the beginning of the dictionary.

31 Next, stroke the stroke data to be searched [1
- Take out from Kubasov 77.

32-37 The dictionary data indicated by the search pointer for each character type is compared with the stroke data. If correct, move on to the next stroke comparison; if incorrect, advance the search pointer until the correct condition is satisfied.

38. After the stroke is completed, store the character code corresponding to the point indicated by each pointer 14a to 14c in each character type buffer 10a to 10c.

FIG. 4 is a flowchart showing the processing procedure in the output unit 5.

The corresponding character code is stored in the output buffer according to the value of the character type designation buffer 11. If the value of the specified character buffer is an unrecognized code, the unrecognized code will be output.

FIG. 5 shows the hardware of one embodiment of the present invention (1 is the origin. The entire control is performed by a microcomputer 40, and each processing program and dictionary for recognition is stored in a ROM 41.
and various buffers are set in the RAM 42.

The recognition result is output from an output terminal 45 via an output control circuit 44.

〔Effect of the invention〕

As explained above, in the present invention, there is no need to specify the character type before inputting characters (there is no interruption of thinking due to changing the character type.Moreover, if you enter the wrong character type for a character once, input the character type specification key) It is easy to realize a system that does not require inputting characters again by re-entering the characters.At this time, there is no need to perform recognition processing internally again due to the change of character type (simply changing the buffer corresponding to the character type specification code). The advantage is that the time required for processing l1jj is very short since it is only necessary to output the contents.

When the device of the present invention is realized as described above, it is possible to provide a recognition device that is easy to use without requiring any special hardware device.

Note that the character types to be recognized are not limited to those in this embodiment.

[Brief explanation of drawings]

FIG. 1 is a software configuration diagram of the present invention. Figure 2 is ji+'! Search part data (1 is the yakuchi. Figure 3 is a flow diagram showing the procedure of jikoshusaku. Figure 4 is a flowchart showing the procedure of the output part. Figure 5 is the hard fair (1 is the Cause. 10a to 10c...Character code buffer 11.
... Character type specification buffer 13 ... Output buffer 7 14a to 14c ... Dictionary search blank 40.
...Microcomputer 41...RO
M 42...RAM 43...Tablet control circuit 44...Output control circuit and above Applicant Seiko Epson Co., Ltd. Agent Patent attorney Tsutomu Mogami and 1 other person 3 Figure

Claims (1)

[Claims] When inputting handwritten characters, there is an input section that detects the two-dimensional coordinates of the pen point and identifies the shape of the stroke determined by the on/off state of the pen point, the character code and the standard strokes that make up the character. , a character dictionary hierarchically configured by a combination of segments defined by a plurality of stroke sequences, and a recognition unit that recognizes input characters by referring to the input strokes and the character dictionary. , the recognition unit has as many dictionary search pointers and result storage buffers as the number of character types to be recognized, such as English letters, numbers, kanji, hiragana, etc. Search the dictionary for the character type according to the dictionary search pointer, store the character code of the character type for which a result is obtained in the result storage buffer, and after detecting the end of one character input, specify the character type to be output. An online handwritten character recognition device characterized in that a character code to be finally output is determined by a character type determining means.