JPH042990B2 - - Google Patents

Description

Detailed Description of the Invention (Field of Industrial Application) The present invention relates to an online character recognition method in an online character recognition device used as an input device of an information processing machine, and particularly relates to character frames preset on a tablet. Conventionally, when an online character recognition device recognizes input characters written on a tablet, the characters are sequentially output from the tablet. Detection of information on how many strokes constitute one character based on the stroke information of an input character (hereinafter referred to as character extraction processing) has been performed using the following method.

In other words, when an input character is written on the tablet, the stroke made up of the coordinate data output from the tablet as the pen moves from pen-on to pen-off will move into which character frame among the character frames set on the tablet in advance. The character is extracted based on the information as to whether it belongs to the character. In this case, when strokes belonging to different character frames are output from the tablet, the strokes belonging to the same character frame up to that point are regarded as strokes constituting one character, and character extraction processing is performed.

(Problems to be solved by the invention) However, according to the above-mentioned prior art system,
A character written outside the character frame is output from the tablet as stroke information spanning multiple character frames, even though it is only one character.
As a result, there was a serious drawback in that the incorrect character was cut out, and the input character that was written was separated into a plurality of characters, resulting in erroneous recognition.

The present invention has been made in order to eliminate the drawbacks of the prior art described above, and provides an online character recognition device that can perform correct recognition processing even when input characters are written at any position regardless of the character frame. The purpose is to

(Means for Solving the Problems) The present invention has a tablet that outputs position information of an input character when writing a character as X, Y coordinate point sequence data, and the X, Y coordinates output from the tablet. This is an online character recognition method for an online character recognition device that recognizes input characters written based on coordinate point sequence data, and recognizes input characters written without regard to the character frame set in advance on the tablet. , a stroke data processing section,
It includes a character width calculation section, a character cutout processing section, and a character position calculation section. The stroke data processing section inputs X and Y coordinate point sequence data from the tablet, sequentially extracts stroke information as input characters are written, and outputs the extracted stroke information to the character width calculation section. The character width calculation section inputs stroke information from the stroke data processing section, calculates character width information for each stroke information input, and outputs the calculation result to the character cutout processing section. The character segmentation processing unit performs character segmentation processing by comparing the character width information sequentially output from the character width calculation unit with a predetermined upper limit value of the character width of the input character. Here, the character width upper limit value is stored, for example, in a register or the like, so as to define the maximum permissible size of an input character to be written. The character position calculating section calculates character position information of the characters cut out by the character cutting section.

(Operation) According to the present invention, since the online character recognition system is configured as described above, each technical means operates as follows.

The stroke data processing section extracts stroke information such as the maximum and minimum values of the X and Y coordinates of the stroke based on the X and Y coordinate point sequence data of the input character from the tablet. Each time the extracted stroke information is sent, the character width calculation section calculates character width information including previous stroke information. The character cutting processing unit compares the calculated character width with the upper limit value of character width corresponding to the upper limit value of the size of the input character to be written, and if the former exceeds the latter value, the character cutout processing unit composes the stroke up to the previous stroke. The character extracted is regarded as one character and the character is extracted, and the character position calculation section calculates the position information of the extracted character so that the recognition process can be performed. Therefore, it is possible to recognize input characters written on the tablet without relying on character frames set in advance, and the problems of the prior art described above can be solved.

(Embodiment) Fig. 1 is a block diagram showing an embodiment of the present invention, in which numeral 1 is a tablet for inputting data of written characters, numeral 2 is a tablet for inputting coordinate data from tablet 1, noise removal and smoothing. , a preprocessing unit that performs processing such as normalization, 3 a stroke data processing unit that extracts stroke characteristic information, stroke position information, etc. for each stroke of a written character, 4 outputting stroke information from the stroke data processing unit 3. 5 is a character width calculation unit that calculates the character width for each character width calculation unit; 5 is a character extraction processing unit that performs character extraction processing using the character width information output from the character width calculation unit 4;
is a character position calculation unit that calculates character position information of characters cut out by the character cutout processing unit 5; 7 is a character dictionary in which stroke characteristic information of characters is stored in advance; 8 is extracted from the stroke data processing unit 3; This recognition unit performs recognition processing by comparing and collating the stroke characteristic information of the characters extracted by the character extraction processing unit 5 and the stroke information of the characters stored in the character dictionary 7.

Next, the operation of this embodiment will be explained. The X and Y coordinate point sequence data in chronological order of the written input characters output from the tablet 1 is sent to the preprocessing unit 2, where it undergoes known preprocessing such as noise removal, smoothing, and normalization. After that, the data is sent to the stroke data processing section 3. In the stroke data processing section 3,
From the stroke coordinate data of the stroke S ₁ (i=1, 2...) from pen-on to pen-off, which is sequentially output as input characters are written, shape characteristic information of the stroke; maximum and minimum values of the X and Y coordinates of the stroke, Stroke position information and the like indicated by x _naxi , y _naxi , x _nioi , y _nioi (i=1, 2...) are extracted. In the character width calculation unit 4, each time the position information of the stroke S _i described above is output from the stroke data processing unit 3, the character width calculation unit 4 calculates the position information from S ₁ to S i including the previous strokes from S ₁ to S _{i -1.} From the above stroke position information, the maximum of x _naxi and y _naxi is determined as X _naxi and Y _naxi
And the smallest of x _nioi and y _nioi is X _nioi and Y _nioi
At the same time, the X and Y direction character width MX _i of the character composed of strokes from S ₁ to S _i ,
MY _i is calculated according to the following formula, and the results of these calculations are output to the character segmentation processing section 5.

MX _i =X _naxi −X _nioi MY _naxi −Y _nioi (i=1, 2) In the character extraction processing unit 5, the character width MX _i ,
Every time MY _i is output, the X and Y direction character width upper limit values P _nx , P _ny and MX _i , which are preset in registers etc. to define the maximum allowable dimensions of input characters to be written, are
MY _i is compared, and when MX _i >P _nx or MY _i >P _ny , the previous stroke up to S _i-1 is regarded as one character, and the character is extracted. The character position calculation unit 6 calculates the center coordinate value of the character cut out by the character cutout processing unit 5, and calculates the character position in the row, column, etc. of the character frame set in advance on the tablet corresponding to this center coordinate value. Calculate information. The recognition processing unit 8 uses a technique such as a known stroke analysis method to combine the stroke feature information of the input character extracted as described above with the stroke information of the character dictionary 7 in which stroke feature information for each character is stored in advance. Compare and match and perform recognition processing.

Next, an example in which the kana "mo" and "ya" are written is shown in FIG. 2, and character extraction processing in this case will be explained. In the figure, 9 to 13 are the strokes when writing the kana "mo" and "ya", and it is assumed that the strokes are written in the order of these numbers. Also, 14 is the X direction character width MX ₄ , 15 is the Y direction character width MY ₄ , 16 is the X direction character width upper limit P _nx , 17
is the Y-direction character width upper limit value _Pny .

The relationship between the character widths MX _i , MY _i (i=1, 2, 3) and the preset character width upper limit values P _nx , P _ny when strokes 9 to 13 are written is MX _i >
P _nx , MY _i < P _ny . When the first stroke 12 of “ya” is written, the character width
The relationship between MX ₄ , MY ₄ and character width upper limit values P _nx and P _ny is
MX ₄ > P _nx and MY ₄ < P _ny . Here, up to stroke 11 immediately before stroke 12 is written is considered as one character, and the character "mo" composed of strokes 9 to 11 is cut out, and stroke 12 is used as the first stroke of the next character "ya". Thereafter, as described above, each time a stroke is input, MX _i and MY _i are compared with P _nx and P _ny to perform character extraction processing.

(Effects of the Invention) As explained above, the present invention uses character width information of input characters without relying on character frame information set on the tablet in advance to perform character cutting processing. Even when input characters are written outside the frame, character extraction processing can be performed correctly. Therefore, according to the present invention, the recognition rate can be improved, and the burden on the scribe is reduced because the scribe can write input characters in any position without being restricted by the character frame set on the tablet. This has the effect of not only significantly reducing the amount of energy used, but also greatly improving the utilization efficiency of the device.

[Brief explanation of drawings]

FIG. 1 is a block diagram of an embodiment of the present invention, and FIG. 2 is a diagram illustrating character extraction processing of input characters. 1... Tablet, 2... Preprocessing section, 3... Stroke data processing section, 4... Character width calculation section, 5
...Character cutting section, 6...Character position calculation section, 7
...Character dictionary, 8...Recognition processing unit, 9-13...
Input character stroke, 14...X direction character width
MX ₄ , 15...Y direction character width MY ₄ , 16...X
Direction upper limit value P _nx , 17...Y direction upper limit value P _ny .

Claims (1)

[Claims] 1. When writing a character, it has a tablet that outputs the positional information of the input character as X, Y coordinate point sequence data, and recognizes the written input character based on the X, Y coordinate point sequence data output from the tablet. An online character recognition method for an online character recognition device, the stroke data processing unit receiving data from the tablet, sequentially extracting and outputting stroke information as input characters are written, and the stroke data processing unit. A character width calculation unit that receives stroke information from and calculates character width information each time stroke information is input, and specifies the character width information sequentially output from the character width calculation unit and the maximum allowable dimension of input characters to be written. In order to do this, the present invention includes a character cutting processing unit that performs character cutting processing by comparing a character width upper limit value of an input character set in advance, and a character position calculation unit that calculates character position information of the cut out character, An online character recognition method characterized by recognizing input characters written in writing regardless of a character frame set in advance on a tablet.