JPH053633B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION <Technical Field> The present invention relates to a handwritten character recognition device.

<Prior Art> Generally, in handwritten character recognition devices, grouping is unavoidable in order to recognize similar characters as intended by the scribe. For example, the kanji ``chikara'' and the katakana ``ka'' may be written in the same way but have completely different meanings, or the kanji ``Z'' may be written as the kanji ``otsu'' or the hiragana ``ka'' depending on how it is written. In cases where the number ``2'' or the alphabetic letter ``Z'' corresponds to four types of characters, group them in advance by ``kanji'', ``hiragana'',
By classifying ``Katakana'', ``Numbers'', ``Alphabets'', ``Symbols'', etc., and specifying one group among them, more accurate recognition can be achieved.
Therefore, the more detailed the grouping is, the higher the recognition rate will be.

However, the subdivision of grouping means an increase in the number of group specifying operations that the scribe must perform before entering characters, which leads to an extreme decrease in operability. The fewer the number of operations (specifying operations for grouping), the better the usability.

This will be illustrated with an example sentence.

Example sentence: I prefer the Japanese-made B747 to the American-made F15 camera.

The grouping designation is indicated by ( ), and the writing is indicated by □.

(a) When grouping is "Kanji", "Hiragana", "Katakana", "Number", "Alphabet", "Symbol" (Kanji) □ I (Hiragana) □ wa (Symbol) □, (Katakana) □ American □ American □ Li □ Ka (Kanji) □ Made (Hiragana) □ (English characters)
□F (number) □1 □5 (katakana) □ □ □ □ □ □ □ □ □ □ □ □ □ (symbol) □, (kanji) □ Japanese □ Book □ Made (hiragana) □ (English characters) □ B ( Number) □7□4□7 (Hiragana)□ (Kanji)
□ 方 (hiragana) □ ga (kanji) □ ko (hiragana) □ ki □ de □su (
symbol) □.

Therefore, the group specification for 34 written characters is
This will be the 21st time.

(b) Grouping into "Kanji/Hiragana", "Katakana",
When using "alphanumeric characters" and "symbols" (Kanji, hiragana) □I□ is (symbol)□, (katakana)
□America□Li□Kanji (kanji/hiragana)□Made□ (alphanumeric characters)
□F□１
□5 (Katakana) □ Card □ Mail □ (Kanji/Hiragana) □ Yo □ Ri (Symbol) □, (Kanji/Hiragana) □ Japanese □ Book □ Made □ (alphanumeric)
□B
□7□4□7 (kanji/hiragana) □'s □ way□ is □like□□
At □ (symbol) □.

Therefore, the group specification for 34 written characters is
12 times, and the number of operations to specify a group has decreased to about 60%.

Based on the above-mentioned background, it has now become common to classify characters into four groups: ``kanji/hiragana,''``katakana,'' ``alphanumeric characters,'' and ``symbols.''

However, among handwritten characters, there are characters (symbols) that frequently appear in sentences and do not have a writing method similar to other characters. That is, punctuation marks such as "," and "."

In the conventional version, even if the above characters (symbols)
Grouping specifications had to be specified one by one, and there was still room for improvement.

<Object of the Invention> Therefore, the present invention provides a handwritten character recognition device equipped with a recognition means that recognizes characters (symbols) that are similar to other characters and have no writing method, regardless of the grouping designation. The purpose is to

<Example> Hereinafter, the present invention will be explained with reference to examples.

FIG. 1 is a diagram showing the external appearance of a handwritten character recognition device according to an embodiment of the present invention, where 1 is a tablet device, and 2 is an input pen having a writing device such as a ballpoint pen at the tip.

This tablet device 1 is, for example, a generally well-known electromagnetic induction type coordinate input device, which inputs coordinate points (for example, 10 points/mm) at a predetermined density in advance.
is generated, and by bringing the tip of the input pen 2 close to the tablet surface, the nearby coordinate points Xi and Yi can be obtained.

A curl cord 3 is connected to the input pen 2, and a writing instrument 4 is attached to the tip of the input pen 2, and a stroke switch is turned on and off by pressing the writing instrument 4 against or away from the tablet surface. is built-in.

Therefore, when writing characters on the tablet surface with the input pen 2, the stroke switch is turned ON.
The stroke trajectory written along with the signal appears as coordinate data and is output. 10 on the tablet surface
is a format sheet, and is fitted into a recess corresponding to its thickness. This format sheet 1
The surface of the 0 and the tablet surface are approximately flat.

A white entry form 20 for notes is placed on the top of the format sheet 10.
Characters to be handwritten are entered on the entry form 20.

Further, 5 is a handwritten character recognition control unit, which is connected to the tablet 1 by a curl cord 6. The control unit 5 is connected to an output device (not shown) such as a CRT display device for outputting characters recognized by another curl code 7.

On the format sheet, character entry frames 11, ... having predetermined coordinate values are arranged in a matrix, and on the top surface of these character entry frames 11, ..., there is a fan that is used less frequently when inputting characters, that is, during editing processing. A cushion key frame 14 is arranged in the horizontal direction.

Further, on the right side of the character entry frames 11, . . . (on the right side from the viewpoint of operability), a function key frame 13, which is frequently used during editing processing, is vertically arranged.

As described above, the format sheet 10
An entry form 20 having a length corresponding to the width of the character entry frame is placed, and characters to be input are entered in the character entry frame 11 using the writing instrument 4 of the input pen 2. Then, for each filled frame, the locus is taken out as coordinate data and the character is recognized.

FIG. 2 shows an enlarged view of the format sheet that serves as the input surface.

Reference numeral 11 denotes character entry frames, which are arranged in a matrix on the format sheet as described above, and each frame is formed into a square of, for example, 10 mm x 10 mm. Also, 13A, 13B, 13C, 13
D is a key for instructing grouping of input characters, 13A indicates kanji/hiragana, 13B indicates katakana, 13C indicates alphanumeric characters, 13D
gives the symbol instructions.

Next, FIG. 3 shows a block diagram of this embodiment, and the configuration will be described in detail.

In the figure, 20 is a CPU (central processing unit) that recognizes the corresponding character based on the stroke data input from the tablet and performs sequence control of the process up to output, and the system program is stored in advance in the ROM 21. controlled by Applicable
The ROM 21 stores system programs classified into five categories. Each program is as follows. That is, the preprocessing program 31 performs a smoothing process to remove stair-like or spike-like trajectory distortions in the locus of stroke data that is sequentially inputted from the tablet in chronological order, and performs interpolation between the coordinate points input and captured from the tablet. Interpolation processing, sampling processing to capture the necessary amount of each coordinate data input from the tablet, normalization processing to reduce/enlarge and normalize the input stroke position and size, and input character input. It is for completion processing. Stroke feature point extraction process 32
is a processing program that uses arithmetic processing to calculate the start point, end point, midpoint, etc. of a stroke from the stroke data obtained by the above-described processing, and extracts them as feature points of the stroke. The distance calculation program 33 compares the above-mentioned feature points with a standard pattern (a combination of a standard pattern group according to grouping instructions and a standard pattern group unrelated to the grouping instructions) 23 to be described later, and calculates the difference between the two. Calculate the distance between and calculate the distance data. The recognition processing program 34 selects the most suitable character based on the distance data and selects it as a recognized character. The output processing program 35 outputs at least one character (JIS
This is for performing output processing to send out the code) to an output device (for example, a CRT display device). Each of the above programs is processed in a time-division manner, and corresponding characters are recognized from sequentially input stroke data.

A data memory 22 is composed of a RAM, and is provided with various buffers (Buf) as described below. Reference numeral 41 denotes a coordinate data buffer that temporarily stores coordinate data input from a tablet and performs preprocessing on the data. Reference numeral 43 denotes a buffer for storing feature points such as the start point, end point, midpoint, etc. of the stroke from the stroke data obtained by the above-mentioned preprocessing. Reference numeral 44 denotes a result buffer that matches the standard pattern based on the data of the feature point buffer, stores the JIS code of the corresponding recognized character in the lowest order as a result of distance calculation, and stores it together with the distance data. In addition to the above-mentioned buffers, an area 42 is provided which includes various buffers for each of the above-mentioned processes and a buffer for storing instructions for grouping, which is a feature of the present invention.

A memory 23 stores the standard pattern, and stores the JIS code of each character as well as standard feature points such as the start point, end point, and midpoint of the character for each stroke. The standard pattern 23
consists of a pattern 51 unrelated to grouping, a pattern 52 for a kanji and hiragana group, a pattern 53 for a katakana group, a pattern 54 for an alphanumeric group, and a pattern 55 for a symbol group.

Note that 24 is an interface device, which is connected to a tablet 25 and an output device 26.

Next, the operation of the circuit configuration shown in FIG. 3 will be explained with reference to the operation flowchart shown in FIG. 4.

First, when the handwritten character input mode is set using the function key or the like, the operation is as shown in FIG. 4. That is, the coordinate data is imported from the tablet, and the coordinate data (Xi, Yi) of the position where the input pen is close to the writing surface or function key surface of the tablet is input via the interface 24, and the coordinate data is stored in the coordinate data buffer 41. is memorized. Also, at this time, it is determined whether the stroke switch at the tip of the input pen is ON or not, and if it is OFF, the stored data is not used and is awaited until it is turned ON. Then, when the stroke switch is turned on, that is, when a stroke start point is input, its coordinate data is taken in, and it is determined whether the input position is a writing frame or a function key based on the value of the coordinate data (steps 100, 101, 102, 103).

In the case of a function key frame, if it is a grouping switching key, designated group information is stored in the various processing buffers 42 (steps 140, 141).
In this case, in order to provide the feeling of operation as a key, once the stroke switch ON is detected in the function key frame, and the processing of the specified function key is completed, the next processing will not be performed until the stroke switch is turned OFF ( Step 142,
143).

In the case of a character entry frame, strokes corresponding to characters input by the operator are continuously input using this first coordinate point as a starting point. That is, the coordinate point next to the starting point is taken in, and the frame position and coordinate values within the frame are determined in the same manner as above (steps 104, 105, 106). Furthermore, smoothing processing, sampling processing, and interpolation processing are performed based on the coordinate values between the two points (step 108). Stroke data is input by repeating this operation.

When one stroke input is completed, the input pen is removed from the writing surface of the tablet, and the stroke switch is turned OFF. If this condition is determined (step 105), the stroke is deemed complete and the process moves to step 120.

In step 120, the number of strokes is counted (corresponding to the number of strokes), the timer is turned on for a few seconds (step 121), and if the timer is left as it is until the timer is turned off (steps 122, 123, 124), the input is considered complete. Move to step 130. Also, if the next stroke start input is received, the timer will start.
Turn it off (step 125) and perform stroke input processing again.

If a new stroke input occurs in another frame during stroke input, it is determined in step 107, the input is considered complete, and the process moves to step 130.

Once the input is completed, the process moves on to the recognition process.The first step in the recognition process is to select a candidate character by comparing it with a standard pattern to see if it is a character that should be recognized regardless of the grouping specification ( Step 130).

Next, the necessity of grouping is determined (step 131), and candidate characters are selected by comparing with the standard pattern for each group (steps 132, 133, 134,
135).

Among the candidate characters, the one with the smallest distance calculated during matching is determined to be the recognized character and the result is output. Furthermore, if an input is made to another frame, the above stroke input process is executed again (step 136,
137).

As mentioned above, characters are recognized by comparing the standard pattern corresponding to the function key operation with the input handwritten character, but the kanji/hiragana group pattern, katakana group pattern, and English If it corresponds to a character (symbol) that does not correspond to a number group pattern or a symbol group pattern, "," or ".", the corresponding character (symbol)
Recognized as ",", ".". That is, the characters (symbols ",", "." are collated and recognized as the characters (symbols) ",", "." in the pattern memory 51 of the standard pattern 23 regardless of which function key is used. be.

Character input according to the present embodiment is shown in correspondence with example sentences.

Note that ( ) indicates grouping designation, and □ indicates writing.

(Kanji/Hiragana) □I□は□, (Katakana)□A□Me□
□Ka (Kanji/Hiragana) □Made□’s (Alphanumeric) □F□1□5 (Katakana)
□Kara□Mea□Ra (Kanji/Hiragana)□Yo□ri□, □Japanese□Book□
Made □ (alphanumeric) □ B 7 □ 4 □ 7 (kanji/hiragana) □'s □ way □ is □
I like □ki□ and □do□.

Therefore, there are 8 group specifications for 34 written characters.
decreases in times.

<Effects> As described above, according to the handwritten character recognition device of the present invention, for characters that have no other similar characters,
The operation of specifying the type of characters (specifying grouping) is no longer necessary, the number of operations for specifying the type of characters for handwritten characters is reduced, the time required for input is shortened, and the operation can be improved.

[Brief explanation of the drawing]

Fig. 1 is a perspective view showing the appearance of the device of the present invention;
3 is a block diagram showing the control circuit configuration of the device, and FIG. 4 is a flowchart showing the operation of the configuration shown in FIG. 3. 23...Standard pattern storage memory, 51...Pattern memory unrelated to grouping, 52...Pattern memory for Kanji and Hiragana groups, 53...Pattern memory for Katakana groups, 54...Pattern memory for alphanumeric groups, 55...Patterns for symbol groups memory.

Claims (1)

[Scope of Claims] 1. Designation means for designating grouping of character types; Storage means for storing the character types designated by the designation means; and Storage means for storing the character types stored in the storage means. Accordingly, a first recognition means recognizes a handwritten input character, and a second recognition means recognizes a character for which there is no other similar character regardless of the type of the specified character. A handwritten character recognition device comprising: