JPH07325888A - Dictionary registering method, character recognizing method and character recognizing device - Google Patents

Abstract

PURPOSE:To rightly recognize a character inputted by the order of making strokes which is partially different from that at the time of registration even if the character is registered by a kind of the order of making-strokes. CONSTITUTION:A stroke selection part 22 extracts the crown and the right-hand radical parts registered in an order of making strokes dictionary 43 from the stroke information stored in the stroke information buffer 311 by a stroke information input part 21. An order of making strokes addition part 23 reads the order of making strokes corresponding to the extracted crown and the right-hand radical parts from an order of making strokes sequence dictionary and adds the order to the order of making strokes at the time of inputting handwritten characters. A registration means 24 registers the orders of making strokes with stroke information in an individual dictionary 42. When a character recognition is performed for the stroke information inputted separately in the stroke information buffer 311, an order of making strokes reconstitution means 25 extends this stroke information by each order of making strokes registered in the individual dictionary 42 corresponding to this stroke information. A character recognition means 26 performs a character recognition for newly inputted stroke information for the stroke information extended by each order of making strokes.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a dictionary registration method when a user registers handwritten characters in a dictionary used for online handwritten character recognition, a character recognition method when this registration method is adopted, and a character adopted by these methods. Regarding the recognition device.

[0002]

2. Description of the Related Art Recently, electronic notebooks, word processors and the like having handwritten character recognition have been commercialized. In these information processing apparatuses, handwriting information such as characters drawn by a user on a transparent tablet with a pen is captured as coordinate value sequence information by the transparent tablet, and the coordinate value sequence information is registered in advance in a recognition dictionary as stroke information. The stroke information is recognized as a character by collating it with the existing character information, and this is input into the device. However, handwritten characters may not be correctly recognized by all users so far, because there are cases where the input person has a habit of writing and the user sometimes writes with his or her own abbreviation. As a means to solve this, the user registers a character he / she writes, creates a personal dictionary, and recognizes the character by comparing the stroke information fetched from the transparent tablet with the registered information in the personal dictionary as described above. As a result, the recognition rate of handwritten characters is improved.

However, most of the character recognition systems of the above-mentioned devices equipped with a personal dictionary in which the user can register his / her written characters are subject to the correct stroke order. Therefore, even characters registered by the user cannot be correctly recognized unless they are entered in the same stroke order as when they were registered. By the way, the same character written by the same person is almost always written in a fixed stroke order, but in some cases, the stroke order may be partially changed. For example, when writing a character that includes "shinnyo", there are both when writing a dot and then the long part, and when writing the long part and then the dot. Such a stroke order variation can be dealt with by pre-registering it in the standard dictionary.

[0004]

On the other hand, when the user registers by himself / herself, it is necessary to register all the patterns that may be written by himself / herself, which requires time and labor for registration. It takes too much time and becomes impractical, and if the user registers the possible pattern in the dark cloud, the capacity of the personal dictionary will be eaten up by that much, and the number of registered characters will decrease. There was a flaw.

The present invention has been made in view of the above circumstances, and by simply registering handwritten characters in a personal dictionary in a single stroke order, it is possible to simultaneously register a plurality of stroke orders related to the characters. Method, character recognition method that can correctly recognize handwritten characters that are input in a stroke order that is partially different from that at the time of registration, and by simply registering handwritten characters in a personal stroke order in a personal dictionary An object of the present invention is to provide a character recognition device that can be registered at the same time and that can correctly recognize handwritten characters input in a stroke order that is partly different from the registration order.

[0006]

According to the invention of claim 1, there is provided a dictionary registration method for registering new handwritten character information in a dictionary to be referred when recognizing handwritten character information input from a coordinate input device. Therefore, a stroke order dictionary in which a plurality of stroke orders for all or a part of the character information forming one character is registered in advance is created, and new handwritten character information is input from the coordinate input device. And, all the stroke order for all or a part of this new handwritten character information is read from the stroke order dictionary, the stroke order at the time of inputting this new handwritten character and the stroke order read from the stroke order dictionary other than this stroke order, It is registered in the dictionary together with new handwritten character information.

According to a second aspect of the present invention, there is provided the dictionary registration method according to the first aspect, in which the stroke order dictionary for all stroke order for all or a part of new handwritten character information input from the coordinate input device is included in the stroke order dictionary. The storage position information of the new handwritten character and the storage position information of the stroke order registered in the stroke order dictionary other than this stroke order and the stroke order at the time of input of this new handwritten character are registered in the dictionary together with the new handwritten character information. To do.

A third aspect of the present invention is a character recognition method for performing character recognition of handwritten character information input from a coordinate input device by referring to a dictionary in which new handwritten characters are registered by the dictionary registration method according to the first aspect. When recognizing the handwritten character input from the coordinate input device, at least one or more stroke order registered in the dictionary corresponding to the handwritten character information and the stroke order at the time of registration of the handwritten character information. so,
After developing the handwritten character information, the handwritten character information is collated with the developed information in each stroke order to perform character recognition.

A fourth aspect of the present invention is a character recognition method for performing character recognition of handwritten character information input from a coordinate input device with reference to a dictionary in which new handwritten characters are registered by the dictionary registration method according to the second aspect. In recognizing a handwritten character input from the coordinate input device, at least one or more stroke order from the stroke order dictionary is stored based on the storage position information registered in the dictionary corresponding to the handwritten character information. Is read out, and after the handwritten character information is expanded in the read stroke order and the stroke order at the time of registration of the handwritten characters in the dictionary, the handwritten character information is collated with the expanded information of each stroke order to perform character recognition. .

According to a fifth aspect of the present invention, in a character recognition device for recognizing character information input from a coordinate input device in the form of coordinate point sequence information by referring to a dictionary, the entire character information forming one character or the character information. When a stroke order dictionary in which a plurality of stroke orders for a part are collected for a plurality of characters and new handwritten character information is input from the coordinate input device, all of the new handwritten character information or all stroke orders for the part thereof is input. Registration means for reading out from the stroke order dictionary, a stroke order at the time of input of this new handwritten character, and the stroke order read by the extraction means other than this stroke order in the dictionary together with the new handwritten character information. Means, at least one or more stroke order registered in the dictionary in correspondence with the handwritten character information input from the coordinate input device, and at the time of registration of the handwritten character information. A stroke order reconstructing unit that develops the input handwritten character information in order, and a character recognition unit that performs character recognition by collating the input handwritten character information with the expanded information developed by the expanding unit. Have a configuration.

According to a sixth aspect of the present invention, in a character recognition device for recognizing character information input from a coordinate input device in the form of coordinate point sequence information by referring to a dictionary, the entire character information forming one character or the whole character information. A stroke order dictionary collecting a plurality of stroke orders for a part of a plurality of characters, and storage position information in the stroke order dictionary of all stroke orders for all or a part of new handwritten character information input from the coordinate input device. A position information search means for obtaining, a stroke order at the time of inputting the new handwritten character, and the storage position information searched by the position information search means other than the stroke order are registered in the dictionary together with the new handwritten character information. Means and at least 1 from the stroke order dictionary based on the storage position information registered in the dictionary corresponding to the handwritten character information input from the coordinate input device. The stroke order is read out, and the stroke order reconstructing means for expanding the handwritten character information in the stroke order read out and the stroke order at the time of registration of the handwritten characters in the dictionary, and the input handwritten character information in the stroke order reconstructing means. And a character recognition means for performing character recognition by collating with the expansion information expanded by the composition means.

[0012]

According to the dictionary registration method of the first aspect of the invention, a plurality of stroke orders for all or part of the character information forming one character are collected for each plurality of characters. When the user inputs new handwritten character information from the coordinate input device, all stroke orders for all or a part of this new handwritten character information are read out from the stroke order dictionary, and when this new handwritten character is input. Since the stroke order and the stroke order read from the stroke order dictionary other than this stroke order are automatically registered in the dictionary together with the new handwritten character information, the user needs to register various stroke orders when registering the handwritten characters. Disappear.

According to the dictionary registration method of the invention of claim 2,
When the user inputs new handwritten character information from the coordinate input device, the storage position information in the stroke order dictionary of all the stroke order of all or part of the new handwritten character information input from the coordinate input device is obtained. , The storage position information of the stroke order registered in the stroke order dictionary other than the stroke order and the stroke order at the time of inputting this new handwritten character is registered in the dictionary together with the new handwritten character information. The user does not need to register various stroke orders when registering characters.

According to the character recognition method of the invention of claim 3,
When recognizing a handwritten character input from the coordinate input device, at least one or more stroke order registered in the dictionary corresponding to the input handwritten character information and the stroke order at the time of registration of the handwritten character information. Since the handwritten character information is developed, the handwritten character information is collated with the developed information in each stroke order for character recognition. Accordingly, when recognizing a handwritten character, various stroke orders at the time of inputting this character are taken into consideration, and the user can perform correct character recognition without writing the character in a fixed stroke order.

In the character recognition method of the invention of claim 4,
When recognizing a handwritten character input from the coordinate input device, at least one or more stroke order is read from the stroke order dictionary based on the storage position information registered in the dictionary corresponding to the handwritten character information. The handwritten character information is expanded in the read stroke order and the stroke order when the handwritten character is registered in the dictionary. As a result, the handwritten character information is collated with the development information in each stroke order for character recognition. Accordingly, when recognizing a handwritten character, various stroke orders at the time of inputting this character are taken into consideration, and the user can perform correct character recognition without writing the character in a fixed stroke order.

In the character recognition device according to the invention of claim 5,
When the new handwritten character information is input from the coordinate input device, the extraction means reads from the stroke order dictionary all the stroke orders for all or part of the new handwritten character information. The registration means registers the stroke order at the time of inputting the new handwritten character and the stroke order other than the stroke order read by the extraction means in the dictionary together with the new handwritten character information. The stroke order reconstructing means includes at least one or more stroke order registered in the dictionary corresponding to the handwritten character information input from the coordinate input device and the input handwritten character information in the stroke order at the time of registration of the handwritten character information. To expand.
The character recognition means collates the input handwritten character information with the expansion information expanded by the expansion means to perform character recognition. This allows user-registered characters to be recognized in consideration of stroke order variations within the same person's writing range, so that you can always be aware of the correct stroke order and register the input pattern in all possible stroke orders at the time of registration. There is no need to do it.

According to a sixth aspect of the present invention, in a character recognition device for recognizing character information input from a coordinate input device in the form of coordinate point sequence information by referring to a dictionary, the stroke order dictionary stores the character information forming one character. A plurality of stroke orders for the whole or a part thereof were collected for each character. The position information retrieving means obtains storage position information in the stroke order dictionary of all stroke orders for all or part of the new handwritten character information input from the coordinate input device. The registration unit registers the stroke order at the time of inputting the new handwritten character and the storage position information searched by the position information search unit other than the stroke order in the dictionary together with the new handwritten character information. The stroke order reconstructing means reads at least one or more stroke order from the stroke order dictionary based on the storage position information registered in the dictionary corresponding to the handwritten character information input from the coordinate input device, and the read stroke order is read. The handwritten character information is expanded in the stroke order and the stroke order at the time of registration of the handwritten character in the dictionary. The character recognition means collates the input handwritten character information with the expansion information expanded by the stroke order reconstruction means to perform character recognition. This allows user-registered characters to be recognized in consideration of stroke order variations within the same person's writing range, so that you can always be aware of the correct stroke order and register the input pattern in all possible stroke orders at the time of registration. There is no need to do it.

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of a character recognition apparatus using the dictionary registration method and the character recognition method of the present invention. 1 is a coordinate input device, 2 is a control device, which is the device body,
Reference numeral 3 is a display device, and 4 is an external storage device. The coordinate input device 1 includes, for example, a transparent tablet and a stylus pen,
It is laminated on a display device 3 such as a liquid crystal display and integrated with the control device 2. The control device 2 includes a stroke information input unit 21, a stroke selection unit 22, a stroke order addition unit 23, a registration unit 24, a stroke order reconstruction unit 25, a character recognition unit 26, an output unit 27, a stroke information buffer 311, and a dictionary. A buffer memory including a stroke buffer 312 and a recognition candidate buffer 313 is provided. The display device 3 is a liquid crystal display or the like as described above, and displays input characters and the like. The external storage device 4 is a magnetic disk device or the like, and stores a standard dictionary 41 to be referred at the time of character recognition, a created personal dictionary 42, a stroke order dictionary 43 for storing strokes to be selected and stroke order thereof, and the like.

Next, the processing of each section in the control device 2 will be described. Stroke information input unit 2 of control device 2
The stroke information buffer 311 receives a coordinate value and pen on / off information from the coordinate input device 1, and treats the stroke from turning on to turning off as one stroke.
Sequentially store in. At this time, when the character recognition means 26 adopts a method of recognizing a combination of preset basic strokes, the stroke information input unit 21 converts the input coordinate information into basic strokes and then stores it in the stroke information buffer 311. . The stroke selection unit 22, the stroke order addition unit 23, and the registration means 24 work at the time of user registration.

That is, the stroke selection unit 22 refers to the stroke order dictionary 43 stored in the external storage device 4, and collects strokes stored in the stroke order dictionary 43 from the stroke group stored in the stroke information buffer 311 ( Select stroke). The stroke order dictionary 43 is shown in FIG.
It is configured as shown in. In the example of FIG.
A stroke group to be selected is written to the left of the selected stroke column, and one stroke is included in the stroke group to the right of the selected stroke column. Further, the stroke order column describes the stroke order within the selected stroke group. Has been done. The pattern information column stores the pattern information of the stroke. Among these, n in the stroke order column represents the maximum number of strokes of the input pattern, and therefore n and n-1 indicate that these strokes correspond to the last two strokes in the input pattern. There is.

Here, the above-described selection stroke extraction can be performed by several methods. The first selection stroke extraction method is a method that uses the relative positional relationship of strokes. For example, when paying attention to "Hen", if "Sanzui", the three strokes written on the left side of the character are extracted as candidates. Also, if it is "kanmuri", a few strokes above the character are extracted as candidates. The extracted candidate stroke is compared with the pattern information stored in the stroke order dictionary 43 in advance to confirm whether or not it is the selected stroke. In the example of FIG. 2, the pattern information is represented by coordinate values, but the form of this information differs depending on the comparison method. The comparison method may be the method performed by the character recognition means 26 or another method.

The second selection stroke extraction method is a method of utilizing the input order. Since "Sanzui" used in the first example is often written first when writing a character, the three strokes at the beginning of writing can be extracted as candidate strokes. Whether or not the stroke extracted as a candidate is the selected stroke can be confirmed by the same processing as the first method. The third selection stroke extraction method is a method of classifying each stroke into a plurality of predetermined basic strokes and extracting a selection stroke candidate by a combination thereof. In this case, whether or not the extracted stroke is the selected stroke may be confirmed by the same method as that described in the first method, and the information (similarity, etc.) when the stroke is classified into basic strokes is used. You can check. The stroke selection unit 22 of this example employs the above-described first selection stroke extraction method.

When a stroke is selected by the stroke selecting unit 22, the stroke order adding unit 23 reads out and interprets the stroke order described in the stroke order dictionary 43 (the stroke order corresponding to the extracted selected stroke). Then, the stroke order information is added to the strokes stored in the stroke information buffer 311. Specifically, the following processing is performed. what if,
It is assumed that the user-registered pattern associated with the character "Kou" is input from the coordinate input device 1 in the stroke order shown in FIG. As shown in FIG. 2, the stroke order dictionary 43 stores three stroke orders for “shinnyo”. That is, the first
As the stroke order corresponding to the stroke order (1, 2), the stroke order shown in (1) of FIG. 4 (which matches the input stroke order in this example) is
As the stroke order corresponding to the second stroke order (n, n-1), the stroke order shown in FIG. (2) (n = 4 in this example) corresponds to the third stroke order (n, n-1). As the stroke order to be performed, the stroke order shown in (3) of the figure is added to the input stroke of the dictionary stroke buffer 311 and stored.

After that, the registration means 24 uses the character recognition means 26.
The input stroke stored in the stroke information buffer 311 and the stroke order information added to the input stroke as well as the character code to be displayed as the recognition result of the character code (corresponding character code in the above example) are associated with each other. The personal dictionary 42 of the external storage device 4 is registered as shown in FIG. However, at the time of this registration, the stroke information buffer 3
Information added in addition to the information stored in 11 depends on the recognition method adopted by the character recognition means 25.

The stroke order reconstructing means 25 and the character recognizing means 26 of the control device 2 work at the time of character recognition. That is, the stroke order reconstructing means 25 operates when the character recognition is performed by referring to the personal dictionary 42, and the registered character pattern stored in the personal dictionary 42 is based on the stroke order information registered corresponding thereto. Then, the strokes are corrected in the order of a plurality of strokes and stored in the dictionary stroke buffer 312. For example, if the registered pattern stored in the personal dictionary 42 is as shown in FIG. 8, the pattern associated with the character "Ai" is the dictionary stroke buffer 31.
As shown in FIG. 5, 2 is stored in 2 in the order of strokes.

The character recognizing means 26 refers to the user registration pattern stored in the dictionary stroke buffer 312 by the stroke order reconstructing means 25 or the dictionary pattern stored in the standard dictionary 41, and strokes stored in the stroke information buffer 311. The information is recognized as a character, and the result is stored in the recognition candidate buffer 313. The character recognition method used here is not particularly limited, and various methods can be used. Finally, the output unit 27 of the control device 2 performs output processing such as displaying necessary information such as input handwriting and recognition result on the display device 3.

Next, the flow of the processing for registering the handwritten character pattern in the personal dictionary by the apparatus shown in FIG. 1 will be described with reference to the flowchart shown in FIG. When the stroke information for one character obtained from the coordinate input device 1 is stored in the stroke information buffer 311 by the stroke information input unit 21 of the control unit 2 in step 601, step 602 is performed.
Then, the stroke selection unit 22 is activated, and it is checked whether or not there is a selected stroke registered in the stroke order dictionary 43 in the input stroke group by the method described above. When it is determined by the above process that the selected stroke exists in the input stroke group, the stroke order adding unit 23 adds the stroke order information to the stroke information buffer 311 in step 603 by the method described above, and then in step 604, the registration means. 24 stores the information in the stroke information buffer 311 in the personal dictionary 42, and ends the processing. However, if it is determined in step 602 that the selected stroke does not exist, the process jumps to step 604, and the stroke information in the stroke information buffer 311 is stored in the personal dictionary 42 only as the stroke order information.

The process of recognizing a newly input handwritten character using the personal dictionary 42 in which the handwritten character of the user is registered is as shown in the flowchart of FIG. First, in step 701, stroke information (handwriting pattern) for one character is input to the stroke information input unit 2
1 is stored in the stroke information buffer 311. Next, the character recognition means 26 determines in step 702 whether or not to use the personal dictionary 42 when recognizing the stroke information. If not, the process proceeds to step 709, and if it is used, the process proceeds to step 703.

In step 703, when the input pattern (stroke group) is recognized by using the personal dictionary 42, first, the index of the personal dictionary pattern is set to 1 (X = 1), and then, In step 704, the stroke order reconstructing unit 25 rearranges the first registered pattern in the personal dictionary 43 in the stroke order stored correspondingly and stores it in the dictionary stroke buffer 312. Next, the character recognition means 26 compares the pattern in the dictionary stroke buffer 312 with the input pattern (stroke information) in step 705 to perform character recognition, and if there is a character that can be recognized as a result, the recognition result. Recognition candidate buffer 3
13 and store the data in step 706. However, if it is determined in step 705 that there is nothing recognizable, the process jumps to step 707.

In step 707, the character recognizing means 26 determines whether or not there is another registered pattern in the personal dictionary 42. If there is, the index is incremented by 1 in step 708, and then the process returns to step 704. The same processing is repeated, and if there is a recognizable one, it is stored in the recognition candidate buffer 313. However, if it is determined in step 707 that another registered pattern does not exist in the personal dictionary 42, the process proceeds to step 710.

On the other hand, in step 709, if the input pattern (stroke group) is recognized by the standard dictionary 41, the character recognizing means 26 performs a predetermined recognition process using the standard dictionary 41 and obtains it. The recognized recognition result is stored in the recognition candidate buffer 313. When the recognition process is completed in this way, the character recognition means 26 rearranges the recognition candidates stored in the recognition candidate buffer 313 in step 710 based on information such as the degree of similarity obtained in the recognition process. In step 711, the output unit 27 performs an output process of displaying the recognition candidate having the highest similarity among the recognition results of the recognition candidate buffer 313 in a predetermined area of the display device 3, and ends the process.

According to this embodiment, the user uses the personal dictionary 42.
If only one way of writing is entered at the time of registration, the device will automatically register the stroke order information of the portion where the stroke order easily changes in the personal dictionary 42, so there is a possibility of writing by yourself. It is not necessary to register all the writing styles, and the user character registration operation can be facilitated. Moreover, since the stroke order information registered at the time of recognizing handwritten characters allows the character recognition in consideration of the stroke order when the stroke order is not exactly the same as when the character registration was performed, the user can recognize the characters. Correct character recognition can be performed without writing in exactly the same stroke order as when registered at the time of recognition, flexibility is provided in the stroke order when handwritten characters are input, and character recognition is performed by registering user characters in the personal dictionary 42. The rate can be improved.

The present invention is not limited to the above embodiment. In the above embodiment, the corresponding stroke order is read out from the stroke order dictionary 43 when the user character is registered, and is registered with the input stroke information in the personal dictionary 42 as shown in FIG.
The personal dictionary 42 stores only the information such as the address of the stroke order dictionary in which the stroke order information to be added is stored, and when the newly input stroke information is recognized as a character, it is stored in the personal dictionary 42 together with the user registration pattern. Based on the stored stroke order information, the actual stroke order is read from the stroke order dictionary 43 shown in FIG. 4, and based on this, the personal dictionary 42 is read.
The same effect can be obtained by adopting the method of expanding in the dictionary stroke buffer 312 in the registered pattern stroke information buffer 311 in the personal dictionary 4 in particular.
The capacity of 2 can be reduced. In addition, although the selection strokes are extracted by some methods in this embodiment, only one of them may be used, or two or more may be used in combination, and another method may be used. May be. In the expansion of the dictionary at the time of recognition, the strokes may be rearranged once before recognition, or the strokes may be sequentially read and recognized while the strokes are rearranged.

[0034]

As described above, according to the dictionary registration method of the present invention, it is possible to simultaneously register a plurality of strokes related to the character just by registering the handwritten character in the personal dictionary in one stroke order. Further, as described above, according to the character recognition method of the present invention, it is possible to correctly recognize handwritten characters input in a stroke order that is partially different from that at the time of registration. Furthermore, as described above, according to the character recognition device of the present invention, by simply registering handwritten characters in the personal dictionary in one stroke order, it is possible to simultaneously register a plurality of stroke orders related to the characters, and at the time of registration. It is possible to correctly recognize handwritten characters that are input in a partially different stroke order.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of an information processing apparatus of the present invention.

FIG. 2 is a diagram showing a configuration example of a stroke order dictionary shown in FIG.

FIG. 3 is a diagram showing registered characters registered in the personal dictionary shown in FIG. 1 and examples of registered patterns corresponding to the registered characters.

FIG. 4 is a diagram showing registered strokes stored in a dictionary stroke buffer 312 shown in FIG. 1 during registration and an example of stroke order information added to the registered strokes.

5 is a diagram showing an example in which a pattern registered in a personal dictionary based on a stroke order added to stroke order information is developed in the dictionary stroke buffer 312 shown in FIG.

6 is a flowchart showing a flow of registration processing when registering a user's handwritten character in the apparatus of FIG.

FIG. 7 is a flowchart showing a flow when handwriting character recognition processing is performed by the device shown in FIG.

8 is a diagram showing an example of the structure of the personal dictionary shown in FIG.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Coordinate input device 2 ... Control device 3 ... Display device 4 ... External storage device 21 ... Stroke information input part 22 ... Stroke selection part 23 ... Stroke order addition part 24 ... Registration means 25 ... Stroke order reconstruction means 26 ... Character recognition means 27 ... Output unit 41 ... Standard dictionary 42 ... Personal dictionary 43 ... Stroke order dictionary 311 ... Stroke information buffer 312 ... Dictionary stroke buffer 313 ... Recognition candidate buffer

Continued Front Page (72) Yoshio Furuichi, 70 Yanagicho, Sachi-ku, Kawasaki-shi, Kanagawa, Toshiba Yanagimachi Plant (72) Inventor, Kazuo Koyama 1381 Shinmachi, Ome-shi, Tokyo 1 Toshiba Computer Engineering Co., Ltd.

Claims (6)

[Claims] 1. A dictionary registration method for registering new handwritten character information in a dictionary that is referred to when recognizing handwritten character information input from a coordinate input device. Or, a stroke order dictionary in which a plurality of stroke orders for a part thereof are registered in advance for each of a plurality of characters is created, and when new handwritten character information is input from the coordinate input device, all of the new handwritten character information or Read all stroke orders for some of them from the stroke order dictionary,
A dictionary registration method characterized in that the stroke order at the time of inputting this new handwritten character and the stroke order read from the stroke order dictionary other than this stroke order are registered in the dictionary together with the new handwritten character information. 2. The dictionary registration method according to claim 1, wherein:
The storage position information in the stroke order dictionary of all stroke orders for all or a part of the new handwritten character information input from the coordinate input device is obtained, and the stroke order at the time of inputting this new handwritten character and other than this stroke order A dictionary registration method characterized in that the storage position information of the stroke order registered in the stroke order dictionary is registered in the dictionary together with the new handwritten character information. 3. A character recognition method for recognizing handwritten character information input from a coordinate input device by referring to a dictionary in which new handwritten characters are registered by the dictionary registration method according to claim 1. When recognizing a handwritten character input from the coordinate input device, at least one or more stroke order registered in the dictionary corresponding to the handwritten character information and a stroke order at the time of registration of the handwritten character information, the handwritten character A character recognition method characterized by performing character recognition by collating the handwritten character information with the developed information in each stroke order after the information is developed. 4. A character recognition method for performing character recognition of handwritten character information input from a coordinate input device by referring to a dictionary in which new handwritten characters are registered by the dictionary registration method according to claim 2. When recognizing a handwritten character input from the coordinate input device, at least one or more stroke order is read from the stroke order dictionary based on the storage position information registered in the dictionary corresponding to the handwritten character information, and the reading is performed. However, the handwriting character information is expanded in the stroke order at the time of registration of the handwritten character in the dictionary and the stroke order, and the character recognition is performed by collating the handwritten character information with the expansion information of each stroke order. Character recognition method. 5. In a character recognition device for recognizing character information input from a coordinate input device in the form of coordinate point sequence information by referring to a dictionary, a plurality of character information forming a character or a part of the character information forming one character is recognized. A stroke order dictionary in which the stroke order is collected for each of a plurality of characters, and when new handwritten character information is input from the coordinate input device, all the stroke orders for all or a part of this new handwritten character information are input from the stroke order dictionary. Extraction means for reading, registration means for registering the stroke order at the time of input of this new handwritten character and the stroke order read by the extraction means other than this stroke order in the dictionary together with the new handwritten character information, and the coordinates. At least one or more stroke order registered in the dictionary corresponding to the handwritten character information input from the input device and the stroke order at the time of registration of the handwritten character information are input. And a character recognition means for recognizing characters by collating the input handwritten character information with the expanded information expanded by the expanding means. Character recognizer. 6. A character recognition device for recognizing character information input from a coordinate input device in the form of coordinate point sequence information by referring to a dictionary, wherein a plurality of character information forming one character or a part of the character information forming one character is recognized. A stroke order dictionary in which the stroke order is collected for each of a plurality of characters, and position information search for obtaining storage position information in the stroke order dictionary of all stroke orders for all or part of new handwritten character information input from the coordinate input device Means, registration means for registering the storage position information retrieved by the position information retrieval means other than the stroke order and the stroke order at the time of input of the new handwritten character in the dictionary together with the new handwritten character information, and the coordinates At least 1 from the stroke order dictionary based on the storage position information registered in the dictionary corresponding to the handwritten character information input from the input device.
A stroke order reconstructing unit that reads out the above stroke order and expands the handwritten character information in the read stroke order and the stroke order at the time of registration of the handwritten characters in the dictionary, and the input unit that expands the input handwritten character information. And a character recognition means for performing character recognition by collating with the expanded information expanded by the character recognition device.