JPH0646419B2 - Method of distinguishing between uppercase and lowercase in handwritten character recognition - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a method for identifying uppercase and lowercase letters in handwritten character recognition.

<Prior Art> A handwritten character recognition device performs character recognition using a tablet that detects coordinate values of a locus of a character written by an input pen. In this handwritten character recognition device, handwritten character recognition is performed by collating a stroke of a character entered in a predetermined frame on the tablet with a standard pattern of the character. Is not necessarily the same size as the standard pattern. Therefore, in order to make it the same size as the standard pattern, expand or reduce the input character stroke,
After that, the handwritten characters are recognized by collating with the standard pattern. However, when all characters are enlarged or reduced to a predetermined size and collation is performed in this way, [Ya-ya],
The uppercase and lowercase letters such as [yu, yu] and [tsu, tsu] cannot be distinguished at all. Therefore, conventionally, a complicated operation of providing a lowercase mode key and pressing the lowercase mode key every time a lowercase character is input is required.

<Objects of the Invention> The present invention has been made in view of the above circumstances, and an object thereof is to accurately recognize characters by enlarging or reducing all handwritten characters to a predetermined size corresponding to a standard pattern. , Uppercase and lowercase letters can be reliably identified without key input, and only the specified key operation can be performed without retyping characters when converting the uppercase letter to lowercase letter or lowercase letter to uppercase letter size after character recognition. Is to provide a method for distinguishing uppercase and lowercase in handwritten character recognition that can be arbitrarily performed.

<Structure of the Invention> In order to achieve the above object, the present invention has a predetermined size in which a pattern of a handwritten character input to a tablet that detects a coordinate value of a locus of a character written by an input pen is predetermined. To the predetermined value by enlarging or reducing the size, and then recognizing the handwritten character by collating with a standard pattern of a predetermined character after obtaining the enlargement ratio or reduction ratio. Whether the handwritten character is uppercase or lowercase is discriminated based on the comparison result of whether it is large or small, and then it is discriminated whether or not there is a key input instructing conversion to uppercase or lowercase. Only when there is a key input to instruct the conversion, the comparison result is converted to lower case when the identification result is upper case, or converted to upper case when the identification result is lower case. It is a method for discriminating uppercase and lowercase letters in handwritten character recognition, which is characterized by outputting as output.

<Example> An example of the present invention will be described below.

FIG. 1 shows an appearance of a handwritten character recognition device to which the present invention is applied.

The tablet surface 1a of the tablet device 1 is formed with a recess corresponding to the thickness of the format sheet 10, and the format sheet 10 is fitted in the recess. The surface of the format sheet 10 and the tablet surface 1a are substantially flat. An entry sheet 12 is placed on the format sheet 10, and characters to be entered are entered on the entry sheet 12.

The input pen 2 uses the curl code 3 and the tablet device 1
The writing instrument 4 is attached to the tip of the input pen 2. The input pen 2 has a built-in stroke switch that is turned on / off by pushing or releasing the writing instrument 4 on the tablet surface 1a. And
When a character or the like is written on the entry form 12 placed on the tablet surface 1a by the input pen 2, the tablet device 1 causes the coordinate data X _i , Y _i representing the locus of the written character or the like in response to the ON signal of the stroke switch. Is output at a density of, for example, 10 points / mm. The tablet device 1 is generally known as, for example, an electromagnetic induction type input device.

.. having a predetermined coordinate value are arranged in a matrix on the format sheet 10. Above the writing frame groups 11, 11, .. At the time of processing, the function key frame 14 which is rarely used is arranged in the horizontal direction. Also, the right side of the input frame group 11, 11, ...
In terms of operability, the right side is better), and the function key frame 13 that is frequently used during the editing process is arranged in the vertical direction. The function key frame 13 is used to perform key input of an uppercase / lowercase conversion key described later.
Character entry frame groups 11, 1 of this format sheet 10
A writing sheet 12 having a length corresponding to the width of 1, ... Is placed, and a writing tool 4 of the input pen 2 is used to write a character to be input in the character frame. In the tablet device 1, the locus is taken out as coordinate data for each written character frame, and the character is recognized.

The control unit 5 is connected to the tablet device 1 by the curl code 6, and the control unit 5 is further connected to an output device such as a CRT display device (not shown) that outputs the character information recognized by the curl code 7.

FIG. 2 shows the configuration of the control unit 5 of the above-mentioned handwritten character recognition device.

The CPU 20 performs sequence control of the process of recognizing the corresponding character based on the stroke data of the handwritten character input from the tablet and outputting this character information.
It is executed according to the system program stored in the ROM 21.

The ROM 21 stores five broadly classified system programs.

The pre-processing program 31 is a smoothing process that removes the stepwise or spike-like trajectory distortion of the trajectory of the stroke data that is sequentially input from the tablet, and the interpolation between the coordinate points of the trajectory that is input and captured from the tablet. Interpolation processing, sampling processing for taking in the necessary amount of each coordinate data input from the tablet, normalization processing for normalizing the position and size of the input stroke by reducing / enlarging, It is composed of a program that performs input completion processing of input characters.

The stroke feature point extraction processing program 32 is a processing program that calculates the start point, end point, midpoint, and the like of the stroke by arithmetic processing from the stroke data obtained by the above-described processing, and extracts it as a feature point of the stroke.

The distance calculation program 33 uses the standard pattern 23 described later.
This is a program for calculating distance data by comparing the above-mentioned characteristic points with the above-mentioned characteristic points and calculating the distance between the two.

The recognition processing program 34 is a processing program that selects the most suitable character based on the distance data and makes it the recognized character.

The output processing program 35 is a processing program for sending at least one character (JIS code) of a plurality of candidate characters to an output device, for example, a CRT display device, by the above-described processes.

As will be described later, the magnification calculation processing program 36 calculates the center of gravity of each character stroke entered in one entry frame 11, and further, uses the center of gravity of each character stroke as the center point to calculate the average value r of the character radius. Is a processing program for calculating upper case / lower case by comparing with a predetermined radius constant rs. Further, the large / small conversion processing program 37 outputs the character code as a result of the recognition processing, and the character code is converted into the character code. In the opposite case, it is a program for processing for converting to a corresponding character code by switching uppercase / lowercase letters using the function keys.

The data memory 22 is composed of RAM and is provided with the following various buffers.

The coordinate data buffer 41 temporarily stores the coordinate data input from the tablet and performs preprocessing on the data. The feature point buffer 43 stores feature points such as the start point, end point, and middle point of the stroke based on the stroke data obtained by the above-described preprocessing. The result buffer 44, based on the data of the feature point buffer 43,
JIS of the recognized character that matches with the standard pattern
As a result of distance calculation, the code is stored in the smallest rank,
Further, it is stored together with the distance data. Data memory 2
In addition, in addition to the above-mentioned buffers, an area 42 formed of various buffers for the above-mentioned processing and the large / large area
The small conversion processing program 37 is provided with a conversion table 45 used for uppercase / lowercase character code conversion.
The conversion table 45 has a conversion table (1) for converting uppercase letters to lowercase letters and a conversion table (2) for converting lowercase letters to uppercase letters.

The standard pattern memory 23 stores standard patterns of characters and the like, and stores, together with the JIS code of each character, standard feature points of the start point, middle point, and end point of each character for each stroke.

Output device 26 such as tablet device 1 and CRT display device
Is connected to the CPU 20 via the interface device 24.

Hereinafter, the operation of the handwritten character recognition device configured as described above will be described with reference to the flowchart of FIG.

First, the coordinate data of the handwritten character on the tablet 1 is fetched (steps S1 and S2), and the preprocessing such as the above-mentioned smoothing process, sampling process, and interpolation process is performed by the preprocessing program 31 (step S3). .

Next, the magnification calculation process of steps S4 to S6 is performed. In this magnification calculation process, the magnification of the input character with respect to the standard pattern is obtained. That is, the number of all coordinate values of the input character stroke is N, and each coordinate value is x _i , y _i (i = 1, ...,
N) is the coordinate value of the center of gravity that is the center point of the input character,
Is calculated by the equations (1) and (2).

Next, the center of gravity, each coordinate value, and the average value r of the distance are calculated by the equation (3).

Where r is a function representing the size of the character,
A circle with a radius r centered at indicates the size of the character. Assuming that the radius of the standard pattern stored in advance in the handwriting character recognition control unit 5 is rs, the scaling factor k is given by
Required by (4).

Here, if k = 1, the standard pattern and the input character have the same size, if k <1, the input character is larger than the standard pattern, and if k> 1, the input character is smaller than the standard pattern.

It goes without saying that the value of the scaling factor k for determining that the input character is a lowercase letter differs depending on the size rs of the standard pattern and the density of the coordinates of the tablet used. For example, with a coordinate density of 10 points / mm (in units of 0.1 mm), the radius of a standard pattern in a 10 mm square character entry frame is rs = 35 (3.5
mm), if k ≥ 1.25 is lower case and k <1.25 is upper case, the input character stroke r ≤ 28 (2.8 mm
If it is less than or equal to), it is determined to be a small letter, and if r> 28 (greater than 2.8 mm), it is determined to be an upper case (steps S7 to S
9).

Next, the input character is enlarged or reduced to a certain size according to the obtained magnification k (step S10). Then, match the enlarged or reduced input characters with the standard pattern,
The input character is recognized (step S11). In this way, the input handwritten character pattern is enlarged or reduced to a predetermined size and the enlarged or reduced character is collated with the predetermined standard pattern of the character. Since the size of the input character becomes the size of the character of the standard pattern, the input character and the standard pattern can be easily compared and collated. At this time, if the lower case flag is on and the character corresponds to the case conversion of the recognition result (steps S12 and S13), the recognition result is converted to a lower case by referring to the conversion table (1) of the data memory 22. (Steps S14, S15). Then, the recognition result is output to the output device 26 (step S1).
6).

Next, the key input of the upper / lower case conversion key provided in the function key frame 13 as a function of editing the recognized character is determined (step S17), and further it is determined whether the recognition result is a lower case (step S18). And
When there is a key input of the uppercase / lowercase conversion key and the recognition result is lowercase, the conversion table (2) of the data memory 22 is referred to and the lowercase character as the recognition result is changed to uppercase (steps S19 and S20). ). Further, if there is a key input of the uppercase / lowercase conversion key and the recognition result is uppercase, the uppercase of the recognition result is changed to lowercase by referring to the conversion table (1) of the data memory 22 (step S21, S22).

In this way, it is possible to convert the character size after recognizing characters from uppercase letters to lowercase letters or from lowercase letters to uppercase letters simply by operating the uppercase / lowercase letters conversion key, so it is necessary to re-enter characters. There is no.

As a method for defining the size of handwritten characters,
As shown in the figure, when a polygon is in contact with the input character and all vertices have one coordinate value of the input character, the area of the polygon is defined as the size of the character and the ratio (magnification) to the area of the standard pattern. It is also possible to determine uppercase and lowercase by using.

<Effects of the Invention> As described above, in the present invention, when the magnification when enlarging or reducing a handwritten character to a certain size is equal to or greater than a predetermined value, the handwritten character is a small letter. Therefore, it is not necessary to perform a key operation each time a lowercase letter is input, and the operation of the handwritten character recognition device can be simplified. Moreover, the character size after character recognition can be arbitrarily converted from uppercase letters to lowercase letters or from lowercase letters to uppercase letters by simply operating the specified keys provided as a function for editing recognized characters. No need to re-fill.

[Brief description of drawings]

FIG. 1 is a perspective view showing the appearance of a handwritten character recognition device, and FIG.
FIG. 4 is a block diagram showing the configuration of the control device of the handwritten character recognition device, FIG. 3 is a flowchart showing the operation of the device of FIG. 2, and FIG. 4 is a diagram showing another embodiment of handwritten character recognition. 1 ... Tablet device, 2 ... Input pen 13 ... Function key frame

Claims (1)

[Claims] 1. A pattern of handwritten characters input to a tablet for detecting coordinate values of a locus of a character entered by an input pen is enlarged or reduced to a predetermined size, and the enlargement ratio or The handwriting character is recognized by collating it with a standard pattern of a predetermined character after obtaining the reduction rate, and the handwriting character is determined by the comparison result of whether the enlargement factor or the reduction rate is larger or smaller than a predetermined value. If uppercase or lowercase is discriminated, then whether or not there is a key input instructing conversion to uppercase or lowercase, and if there is a key input instructing conversion to uppercase or lowercase In the above, the handwritten sentence is characterized in that when the identification result by the comparison is uppercase, it is converted to lowercase, or when the identification result is lowercase, it is converted to uppercase and output. Identification method of uppercase and lowercase letters in the recognition.