JP4269357B2 - Handwritten character recognition device, handwritten character recognition method, and storage medium storing handwritten character recognition program - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a handwritten character recognition device, a handwritten character recognition method for inputting data by handwriting, and a storage medium storing a handwritten character recognition program.
[0002]
[Prior art]
2. Description of the Related Art Generally, some information devices such as electronic notebooks that input and manage arbitrary data include a handwritten character recognition device for recognizing handwritten characters and symbols and inputting them as data.
[0003]
This handwritten character recognition device has a touch panel configured by, for example, overlaying a transparent tablet on a display screen. When characters and symbols are drawn and displayed by handwriting on the touch panel, the handwritten character recognition device is closest to the locus of the drawing. The font pattern of the character or symbol being searched is retrieved from a large number of characters and symbol fonts stored in advance as a dictionary memory and displayed.
[0004]
In such a handwritten character recognition device, when a character having a plurality of similar characters such as the numeral “3” and the hiragana “ro” is input, the “3” and “ro” Are displayed as a list of recognition candidate characters, such as (1 :, 2: 3), and a desired character can be selected and input here.
[0005]
When the second candidate “3” is selected and entered in the list of recognition candidate characters (1: iro, 2: 3), the next time the same handwriting input is made, it is selected last time. The candidate character “3” is replaced with the first candidate and displayed as a list (1: 3, 2: iro).
[0006]
However, when it is desired to alternately input “3” and “ro” as the series of recognition candidate characters, the second candidate character whose candidate order is to be switched must be repeatedly selected each time. Is very troublesome.
[0007]
On the other hand, in order to eliminate the troublesome operation of inputting such confusing characters by candidate selection, for example, each of the four character input areas is limited to input areas for Kanji, Hiragana, Katakana, and alphanumeric characters. In addition, a handwritten character recognition apparatus that recognizes a specific character type in a fixed manner according to an input area is also in practical use.
[0008]
However, in such a handwritten character recognition device in which the input area is limited by the character type, for example, when inputting kanji, the character is input and recognized one by one in the leftmost input area, and when inputting a number. Needs to be entered and recognized one by one in the rightmost input area, and characters cannot be entered in a natural flow from left to right.
[0009]
Furthermore, a handwritten character recognition device that combines a plurality of input areas that do not limit the character type of input and fixed input areas for alphabets and numbers is also considered. In this case, however, alphabets and numbers are used as fixed input areas. Therefore, there is a problem that it is difficult to input characters in a natural flow from left to right as described above.
[0010]
[Problems to be solved by the invention]
As described above, the conventional handwritten character recognition device eliminates the need to perform a confusing character candidate selection operation between different character types, but since the input area is fixed for each character type, it is natural from left to right. There is a problem that characters cannot be input in the flow.
[0011]
The present invention has been made in view of the above problems, and allows natural character input from the left to the right without fixing the input area for each character type, and is confusing to exist between different character types. An object of the present invention is to provide a handwritten character recognition device, a handwritten character recognition method, and a storage medium storing a handwritten character recognition program that can correctly recognize characters.
[0012]
[Means for Solving the Problems]
That is, the first handwritten character recognition apparatus according to the present invention has a plurality of handwritten character input areas, and each area has priority. do it Handwritten characters with preset character types Input means and , Within a predetermined time after recognition of the previous input of handwritten characters In the handwritten character input means Continuous input recognition character type setting means for preferentially recognizing the last recognized character type when the current handwritten character is input And an input area recognition character type setting unit for preferentially recognizing the character type set in the area where the handwritten character is input when the current handwritten character is input to the handwritten character input unit at a time other than within the time When, It is characterized by comprising.
[0016]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of the present invention will be described below with reference to the drawings.
(First embodiment)
FIG. 1 is a block diagram showing a configuration of an electronic circuit of a handwritten character recognition apparatus according to an embodiment of the present invention.
[0017]
This handwritten character recognition device includes a control unit 11 using a CPU which is a computer.
The control unit 11 includes a key input signal from the key input unit 12, a handwritten input signal from the handwriting input unit 12 a, and a touch panel 15 configured by superimposing a transparent tablet 14 on the handwritten character display area 13 </ b> A in the liquid crystal display unit 13. In response to a handwritten input signal, an external storage medium 18 such as an FD (floppy disk) that is activated by a system program 16a stored in advance in the ROM 16 or read by an external storage reader 17 such as an FDD (floppy disk drive). The program data for handwritten character recognition control stored in is activated, and the operation of each part of the circuit is controlled using the RAM 19 as a work memory.
[0018]
The control unit 11 is connected to the key input unit 12, the handwriting input unit 12a, the liquid crystal display unit 13, the touch panel 15, the ROM 16, the external storage reading unit 17, and the RAM 19, as well as the dictionary data memory 20 and the timer 21. Is done.
[0019]
FIG. 2 is a diagram showing a configuration of a display area in the liquid crystal display unit 13 of the handwritten character recognition apparatus.
The liquid crystal display unit 13 includes a handwritten character display area 13A corresponding to four handwritten character input areas as the touch panel 15, an input / candidate character display area 13B arranged on the handwritten character display area 13A, and the handwritten character display. A kanji character having a priority character type display area 13C arranged on the right of the area 13A, and the four handwritten character input areas in the handwritten character display area 13A preferentially recognize input handwritten characters as kanji characters from the left end. The priority recognition area 13A1, the hiragana priority recognition area 13A2 that is preferentially recognized as hiragana, the katakana priority recognition area 13A3 that is preferentially recognized as katakana, and the alphanumeric priority recognition area 13A4 that is preferentially recognized as alphanumeric characters are set.
[0020]
In addition, the input / candidate character display area 13B includes an input character display area 13B1 in which a handwritten character displayed after handwritten input in the handwritten character display area 13A is recognized and input is displayed, and the handwritten character The candidate character display area 13B2 in which the recognition candidate characters corresponding to are displayed as a list.
[0021]
In the priority character type display area 13C, a priority character type associated with recognition of a handwritten character input and displayed in the handwritten character display area 13A is displayed.
[0022]
The ROM 16 stores in advance a system program 16a that controls the overall operation of the handwritten character recognition apparatus, and stores various application programs 16b and a character recognition processing program 16c. Corresponding to the coordinate range associated with each handwritten input, the position of the four areas (Kanji priority recognition area 13A1, Hiragana priority recognition area 13A2, Katakana priority recognition area 13A3, alphanumeric priority recognition area 13A4) in the display area 13A Handwritten BOX position information 16d and the like for attaching and identifying is stored.
[0023]
The RAM 19 includes a system data memory 19a, application data memory 19b, handwritten input coordinate data memory 19c, candidate character similarity data memory 19d, priority character type memory 19e, additional numerical value memory 19f, continuous character input timer memory 19g, continuous character input time-out. A memory 19h and a work memory 19i are provided.
[0024]
In the system data memory 19a, when the system program 16a stored in the ROM 16 is activated, the system program 16a is once read and stored in the system data memory 19a.
[0025]
In the application data memory 19b, when an application program corresponding to an execution function among the various application programs 16b stored in the ROM 16 is started, the application program is once read and stored in the application data memory 19b. Remembered.
[0026]
FIG. 3 is a diagram showing a data storage state in the handwritten input coordinate data memory 19c provided in the RAM 19 of the handwritten character recognition apparatus.
In the handwritten input coordinate data memory 19c, a drawing trajectory accompanying the input of the handwritten characters on the touch panel 15 is displayed in the handwritten character display area 13A where the tablet 14 detects the handwritten strokes at regular time intervals. Corresponding x and y coordinate data is stored.
[0027]
FIG. 4 is a diagram showing a data storage state in the candidate character similarity data memory 19d provided in the RAM 19 of the handwritten character recognition apparatus.
In the candidate character similarity data memory 19d, candidate characters retrieved from the dictionary data memory 20 in accordance with the character recognition process for handwritten characters input by handwriting on the touch panel 15 are displayed together with the respective character types and similarities. Sorted and stored in descending order. At this time, the area handwritten on the touch panel 15 is one of the four areas (kanji priority recognition area 13A1, hiragana priority recognition area 13A2, katakana priority recognition area 13A3, and alphanumeric priority recognition area 13A4). Is identified according to the handwritten BOX position information 16d stored in advance in the ROM 16, the priority character type is determined, and a preset score value is added to the similarity of the candidate character of the character type matching this.
[0028]
That is, for example, it is identified that the handwritten character that is the source of the candidate character in the candidate character similarity data memory 19d in FIG. 4 is input in the alphanumeric priority recognition area 13A4 of the touch panel 15, and the priority character type is determined as alphanumeric. When the preset score value is “30”, the score “3” and “b”, which are alphanumeric candidate characters, are added to the score “70” and “40”, respectively. “30” is added and changed to “100” and “70” to be recognized with priority (see FIGS. 7 and 8).
[0029]
When inputting handwritten characters on the touch panel 15, the priority character type memory 19e has four input areas (kanji priority recognition area 13A1, hiragana priority recognition area 13A2, katakana priority recognition area 13A3, alphanumeric character priority recognition area 13A4). Is identified according to handwritten BOX position information 16d stored in advance in the ROM 16, and the priority character type is determined and stored.
[0030]
In the additional numerical value memory 19f, the similarity of candidate characters written in the candidate character similarity data memory 19d as a character type that matches the priority character type stored in the priority character type memory 19e should be added for priority recognition. A preset score value (for example, “30”) is stored.
[0031]
In the continuous character input timer memory 19g, the time interval t at which each handwritten character is input with the input of the handwritten character on the touch panel 15 is measured by the timer 21 and stored.
[0032]
In the continuous character input time-out memory 19h, the predetermined time tout for determining whether or not the input time interval t of each handwritten character on the touch panel 15 has exceeded a predetermined time tout meaning a change of the character type is preset. Is remembered. The predetermined time tout of the continuous character input timeout is set to 2 seconds, for example, and it is determined whether or not the priority recognition areas 13A1 to 13A4 are switched according to the change of the character type at the time of continuous character input.
[0033]
In the work memory 19i, data input / output by the control unit (CPU) 11 is temporarily stored and held as necessary in accordance with operation control processing of each part of the circuit by executing various application programs.
[0034]
The dictionary data memory 20 stores in advance kanji, hiragana, katakana, alphabetic characters, numbers, all characters of symbols, pattern data corresponding to each of symbols, and the like.
Next, the operation of the handwritten character recognition apparatus having the above configuration will be described.
[0035]
FIG. 5 is a flowchart showing a character recognition process in the handwritten character recognition apparatus.
FIG. 6 is a diagram showing a display operation of handwritten character recognition accompanying the character recognition processing of the handwritten character recognition device.
[0036]
FIG. 7 shows the display operation of the handwritten character when the alphanumeric character “3”, which is confused with the hiragana character “RO”, is input in the alphanumeric priority recognition area 13A4 in accordance with the character recognition processing of the handwritten character recognition device. FIG.
[0037]
FIG. 8 shows the candidate character similarity data memory 19d in the case where the alphanumeric character “3”, which is confused with the hiragana character “ro” accompanying the character recognition process of the handwritten character recognition device, is input in the alphanumeric character priority recognition area 13A4. It is a figure which shows the similarity memory state of each candidate character.
[0038]
For example, when “Honmachi 3”, which is partial address data, is input by handwriting and displayed for recognition, the character “book” is displayed in the kanji priority recognition area 13A1 of the touch panel 15, as shown in FIG. When handwritten input with the pen, the xy coordinate data corresponding to the handwritten trajectory repeatedly detected at regular time intervals by the tablet 14 accompanying the handwritten input of the handwritten character “book” is sequentially fetched, and handwritten input in the RAM 19 is performed. It is stored in the coordinate data memory 19c (see FIG. 3) (steps S1 → S2, S3 → S2).
[0039]
Here, when the handwriting input for one character of the handwritten character “book” in the kanji priority recognition area 13A1 is completed, it corresponds to the coordinate range of the handwritten character “book” stored in the handwritten input coordinate data memory 19c in the RAM 19. Based on the handwritten BOX position information 16d stored in advance in the ROM 16, it is identified that the handwritten input is made in the kanji priority recognition area 13A1, and the priority character type is determined to be "kanji". It is stored in the priority character type memory 19e in the RAM 19 (steps S3 → S4).
[0040]
Further, based on the coordinate data of the handwritten character “book” stored in the handwritten input coordinate data memory 19c in the RAM 19, all character / symbol pattern data stored in the dictionary data memory 20 are referred to and fixed. A candidate character with a similarity or higher is searched, and the searched candidate character is associated with the similarity and character type and stored in the candidate character similarity data memory 19d in the RAM 19 (step S5).
[0041]
Then, the score value “30” stored in the score value memory 19f is added to the similarity of the candidate character of the character type matching the priority character type “Kanji” stored in the priority character type memory 19e, and the candidate The candidate characters in the character similarity data memory 19d are sorted again in descending order of their similarity and displayed in a list in the candidate character display area 13B2 on the liquid crystal display unit 13 (steps S6 and S7).
[0042]
For the handwritten character “book”, candidate characters below the second candidate are not searched in this case, and only the first candidate character “book” is recognized as shown in FIG. 6B. Is displayed in the input character display area 13B1.
[0043]
Next, when the handwritten character “machi” is input in the hiragana priority recognition area 13A2 that is continuous to the right of the kanji priority recognition area 13A1 that is the input area of the handwritten character “book”, the same character recognition processing as described above is performed. Only the first candidate character “town” is searched and displayed as the recognized character in the input character display area 13B1 as shown in FIG. 6C (steps S1 to S7).
[0044]
Then, as shown in FIG. 6D, the number “3”, which is a character confusing hiragana “ro”, is positioned to the right of the hiragana priority recognition area 13A2, which is the input area of the handwritten character “town”. When handwriting is input in the alphanumeric priority recognition area 13A4, the coordinate data of xy corresponding to the handwritten trajectory repeatedly detected at regular time intervals by the tablet 14 accompanying the handwriting input of the handwritten character “3” is sequentially captured. It is stored in the handwritten input coordinate data memory 19c (see FIG. 3) in the RAM 19 (steps S1 → S2, S3 → S2).
[0045]
When the handwriting input for one character of the handwritten character “3” in the alphanumeric priority recognition area 13A4 is completed, the coordinate range of the handwritten character “3” stored in the handwritten input coordinate data memory 19c in the RAM 19 is displayed. Accordingly, based on the handwritten BOX position information 16d stored in advance in the ROM 16, it is identified that the handwritten input is made in the alphanumeric priority recognition area 13A4, and the priority character type is “alphanumeric”. It is determined and stored in the priority character type memory 19e in the RAM 19 (steps S3 → S4).
[0046]
Further, based on the coordinate data of the handwritten character “3” stored in the handwritten input coordinate data memory 19 c in the RAM 19, all character / symbol pattern data stored in the dictionary data memory 20 is referred to and fixed. Candidate characters with similarities or higher are searched, and the similarity (80, 70, 50, 45, 40) and character type (Hiragana / English) are searched for each candidate character (ro, 3, o, u, b). Numbers, kanji, hiragana, and alphanumeric characters are associated and stored in the candidate character similarity data memory 19d in the RAM 19 as shown in FIG. 8A (step S5).
[0047]
Then, with respect to the similarities “70” and “40” of “3” and “b” which are candidate characters of the character type that matches the priority character type “alphanumeric” stored in the priority character type memory 19e, FIG. As shown in FIG. 8B, the added point value “30” stored in the added point value memory 19f is added, and the similarity of the candidate character “3” is “100”, and the similarity of “b” is As shown in FIG. 8C, the candidate characters in the candidate character similarity data memory 19d are sorted in the descending order of their similarity (3, b, b, end, and so on). The list is displayed again as a list in the candidate character display area 13B2 of the liquid crystal display unit 13, and the first candidate character "3" is displayed as an input character display area 13B1 as a recognized character as shown in FIG. (Step S , S7).
[0048]
Therefore, according to the handwritten character recognition device having the above-described configuration, the input area for four characters for continuously inputting a plurality of handwritten characters is arranged in the left-right direction on the touch panel 15 and is divided by the touch panel 15. The input areas for the handwritten characters are set as the kanji priority recognition area 13A1, the hiragana priority recognition area 13A2, the katakana priority recognition area 13A3, and the alphanumeric priority recognition area 13A4, respectively, and the handwriting input in the input areas 13A1 to 13A4. The similarity of the candidate character retrieved from the dictionary data memory 20 based on the character is given priority by adding a certain additional numerical value to the similarity of the candidate character of the character type that is preferentially recognized in the input area. The touch panel 15 is arranged on the touch panel 15 in order of increasing similarity. In the case of a handwritten character in which confusing characters exist between different character types while continuously inputting handwritten characters by the natural input procedure from left to right in the four character input areas 13A1 to 13A4, priority is given to the corresponding character type. By inputting the misleading handwritten character from the recognition area, the desired character is recognized and displayed as the first candidate character.
[0049]
In the first embodiment, each of the four handwritten character input areas divided by the touch panel 15 is set as a dedicated priority recognition area 13A1 to 13A4 for each character type. As described in the second embodiment, the priority recognition area is set to overlap between character types, and the size of the input area is distinguished between different character types, or the time interval of continuous character input is determined to determine different character types. It is possible to determine which character type of the first handwritten character that is expected to be input from the priority recognition area, and perform priority continuous recognition for each character type while performing natural continuous input of handwritten characters from left to right. It is good also as a structure.
[0050]
(Second Embodiment)
FIG. 9 is a diagram showing a configuration of a handwritten character display area 13A in the case where the priority recognition areas for the respective character types on the touch panel 15 are overlapped and set according to the second embodiment of the handwritten character recognition apparatus.
[0051]
In the second embodiment of the handwritten character recognition apparatus, as shown in FIG. 9, all of the four handwritten character input areas 13A1 to 13A4 are set as kanji priority recognition areas for each character, and the input areas 13A1 to 13A1 are set. 13A4 is divided into halves, and four 1/2 areas from the left end are set as hiragana priority recognition area 13A1L, katakana priority recognition area 13A1R, alphabetic priority recognition area 13A2L, and number / symbol priority recognition area 13A2R, respectively.
[0052]
As described above, in the handwritten character recognition device that constitutes the handwritten character display area 13A of the touch panel 15, when performing continuous input of handwritten characters, priority recognition of the character type set in the area where the first handwritten character is input is performed. This is also applied to handwritten characters that are input continuously within a certain time tout, and when the input time interval t of each handwritten character exceeds the certain time tout, it is determined that the character type has changed, and the handwritten character Is changed as the first handwritten character.
[0053]
FIG. 10 is a flowchart showing a character recognition process of the second embodiment in the handwritten character recognition apparatus.
FIG. 11 is a diagram showing a display operation (part 1) of handwritten character recognition accompanying the character recognition processing of the second embodiment of the handwritten character recognition apparatus.
[0054]
FIG. 12 is a diagram showing a display operation (part 2) of handwritten character recognition accompanying the character recognition processing of the second embodiment of the handwritten character recognition apparatus.
FIG. 13 is a diagram showing a display operation (part 3) for handwritten character recognition accompanying the character recognition processing of the second embodiment of the handwritten character recognition apparatus.
[0055]
When the character recognition process in FIG. 10 is started, first, the handwritten character input time interval t counted by the continuous character input timer memory 19g in the RAM 19 is cleared to "0" (step A1).
[0056]
Then, the time interval t of handwritten character input counted by the continuous character input timer memory 19g is counted up until the handwritten character is input by pen input on the touch panel 15 (step A2 → A3).
[0057]
For example, when “Tsubakigaoka housing complex B5-203”, which is part of the address data, is input by handwriting and displayed for recognition, as shown in FIG. 11A, in the Hiragana priority recognition area 13A1L of the touch panel 15. When the character “tsu” is input by hand using a pen, the time interval t of handwritten character input timed by the continuous character input timer memory 19g is stored in the continuous character input timeout memory 19h for a certain time tout (= 2 seconds), and if it is determined that the continuous character input time interval t exceeds the predetermined time tout, the priority character type stored in the priority character type memory 19e is erased and the priority is set. The priority-recognized character type displayed in the character type display area 13C is deleted (step A2 → A4 → A5).
[0058]
Then, the xy coordinate data corresponding to the handwritten trajectory repeatedly detected at regular time intervals by the tablet 14 in accordance with the handwritten input of the handwritten character “tsu” is sequentially fetched, and the handwritten input coordinate data memory 19c ( (See FIG. 3) (steps A6, A7 → A6).
[0059]
Here, it is determined whether or not the priority character type is stored in the priority character type memory 19e, and the priority character type is determined by the deletion process of the priority character type in step A5 according to the passage of the predetermined time tout in step A4. If it is determined that it is not stored, the handwritten BOX position information 16d stored in advance in the ROM 16 is stored in accordance with the coordinate range of the handwritten character “one” stored in the handwritten input coordinate data memory 19c in the RAM 19. Based on this, it is identified that the handwriting input has been made in the Hiragana priority recognition area 13A1L, and the priority character type is determined to be “Hiragana” and is stored in the priority character type memory 19e in the RAM 19, as shown in FIG. ), It is displayed in the priority character type display area 13C (step A8 → A9).
[0060]
Further, based on the coordinate data of the handwritten character “one” stored in the handwritten input coordinate data memory 19c in the RAM 19, all character / symbol pattern data stored in the dictionary data memory 20 are referred to and fixed. A candidate character with a similarity or higher is retrieved, and the retrieved candidate character is associated with the similarity and character type and stored in the candidate character similarity data memory 19d in the RAM 19 (step A10).
[0061]
Then, the score value “30” stored in the score value memory 19f is added to the similarity of the candidate character of the character type matching the priority character type “Hiragana” stored in the priority character type memory 19e, and the candidate The candidate characters in the character similarity data memory 19d are sorted again in descending order of their similarity and displayed in a list in the candidate character display area 13B2 on the liquid crystal display unit 13 (steps A11 and A12).
[0062]
In this case, for example, “tsu”, “tsu”, and “fu” are searched for candidate characters for the handwritten input character “tsu”. Among the candidate characters, the priority character type “stored in the priority character type memory 19e” is searched. The added numerical value “30” is added to the similarity of “tsu” and “tsu” of the hiragana character type that matches “Hiragana”, so that the correct “tsu” as a handwritten character is the first candidate character, “Tsu” is displayed as a second candidate character, Katakana “fu” is displayed as a third candidate character in the candidate character display area 13B2, and the first candidate character “tsu” is displayed in the input character display area 13B1. Is displayed.
[0063]
When the hiragana handwritten character “tsu” is recognized and displayed in this way, the continuous character input time interval t counted by the continuous character input timer 19g is once reset and restarted (step A13).
[0064]
Here, as shown in FIG. 11 (C), the same character type is used in the Katakana priority recognition area 13A1R, which is adjacent to the right side of the “Hi” of the hiragana priority recognition area 13A1L of the touch panel 15 and subsequently to the right side. When “Hi” of Hiragana is input by handwriting, the time interval t from the end of character input before being restarted and timed by the continuous character input timer memory 19g is stored in the continuous character input time-out memory 19h. It is determined whether or not the predetermined time tout (= 2 seconds) has been exceeded, and when it is determined that the continuous character input time interval t does not exceed the predetermined time tout, the handwritten character “BA” is input by handwriting. Accordingly, the coordinate data of xy corresponding to the handwritten trajectory repeatedly detected at regular time intervals by the tablet 14 is sequentially taken in, and the hand in the RAM 19 is read. Is stored in the can input coordinate data memory 19c (step A2 → A4 → A6, A7).
[0065]
If it is determined that the priority character type “Hiragana” is stored in the priority character type memory 19e, the coordinate data of the handwritten character “BA” stored in the handwritten input coordinate data memory 19c in the RAM 19 is stored. Based on this, all character / symbol pattern data stored in the dictionary data memory 20 are referred to search for candidate characters with a certain similarity or higher, and the similarity and character type correspond to each searched candidate character. The result is stored in the candidate character similarity data memory 19d in the RAM 19 (step A8 → A10).
[0066]
Then, the score value “30” stored in the score value memory 19f is added to the similarity of the candidate character of the character type matching the priority character type “Hiragana” stored in the priority character type memory 19e, and the candidate The candidate characters in the character similarity data memory 19d are sorted again in descending order of their similarity and displayed in a list in the candidate character display area 13B2 on the liquid crystal display unit 13 (steps A11 and A12).
[0067]
In this case, as candidate characters for the handwritten input character “BA”, for example, “BA” and “PA” are searched. Among the candidate characters, the priority character type “Hiragana” stored in the priority character type memory 19e is Since the added point value “30” is added to the similarity between the matching character types of hiragana “ba” and “pa”, as shown in FIG. Are displayed as a list in the candidate character display area 13B2 as the first candidate character and "Pa" is displayed as the second candidate character, and the first candidate character "BA" is displayed in the input character display area 13B1.
[0068]
Thus, when the continuous recognition display of the hiragana handwritten characters “tsu” and “ba” is made, the continuous character input time interval t counted by the continuous character input timer 19g is once reset and restarted (step). A13).
[0069]
Further, as shown in FIG. 11 (D), the hiragana characters “tsu” and “ba” that are continuously handwritten input from the hiragana priority recognition area 13A1L of the touch panel 15 to the right Katakana priority recognition area 13A1R, When the same character type hiragana “ki” is input by handwriting in the adjacent English character priority recognition area 13A2L, the time interval t from the end of character input before being restarted and timed by the continuous character input timer memory 19g. Is determined not to exceed the predetermined time tout (= 2 seconds) stored in the continuous character input time-out memory 19h, and is repeatedly detected at fixed time intervals by the tablet 14 with the handwriting input of the handwritten character “ki”. Xy coordinate data corresponding to the handwritten trajectory is sequentially fetched and the handwritten input coordinate data memory 19c in the RAM 19 is read. (Steps A2 → A4 → A6, A7).
[0070]
If it is determined that the priority character type “Hiragana” is stored in the priority character type memory 19e, the coordinate data of the handwritten character “ki” stored in the handwritten input coordinate data memory 19c in the RAM 19 is stored. Based on this, all character / symbol pattern data stored in the dictionary data memory 20 are referred to search for candidate characters with a certain similarity or higher, and the similarity and character type correspond to each searched candidate character. The result is stored in the candidate character similarity data memory 19d in the RAM 19 (step A8 → A10).
[0071]
Then, the score value “30” stored in the score value memory 19f is added to the similarity of the candidate character of the character type matching the priority character type “Hiragana” stored in the priority character type memory 19e, and the candidate The candidate characters in the character similarity data memory 19d are sorted again in descending order of their similarity and displayed in a list in the candidate character display area 13B2 on the liquid crystal display unit 13 (steps A11 and A12).
[0072]
In this case, for example, “ki”, “mo”, and “ki” are searched as candidate characters for the handwritten input character “ki”. Among the candidate characters, the preferred character type “i” stored in the preferred character type memory 19e is searched. The added numerical value “30” is added to the similarity between “Hi” and “M”, which are the character types that match “Hiragana”, so that it is correct as a handwritten character as shown in FIG. “Ki” is displayed as a list of candidate characters in the candidate character display area 13B2 as the first candidate character, “mo” is the second candidate character, and katakana “ki” is displayed as the third candidate character. “KI” is displayed in the input character display area 13B1.
[0073]
When the continuous recognition display of the hiragana handwritten characters “tsu” “ba” “ki” is made in this way, the continuous character input time interval t counted by the continuous character input timer 19g is once reset and restarted. (Step A13).
[0074]
Then, in order to input by hand the “K”, which is the next character following “Tsubaki” that has been input by handwriting and recognized, the katakana priority recognition area 13A1R on the touch panel 15 is searched for in the continuous character input timer memory 19g. When the time interval t of the continuous character input time counted exceeds the predetermined time tout stored in the continuous character input time-out memory 19h, "K" is input to the katakana priority recognition area 13A1R by handwriting. As shown in FIG. 12A and FIG. 12B, the priority character type “Hiragana” stored in the priority character type memory 19e is erased, and priority recognition up to the previous time displayed in the priority character type display area 13C is performed. Character type “Hiragana” is deleted (steps A 2 → A 4 → A 5).
[0075]
Then, the xy coordinate data corresponding to the handwritten trajectory repeatedly detected at regular time intervals by the tablet 14 in accordance with the handwritten input of the handwritten character “K” is sequentially fetched, and the handwritten input coordinate data memory 19 c in the RAM 19 is loaded. Stored (steps A6, A7 → A6).
[0076]
Here, if it is determined that the priority character type is not stored by the deletion process of the priority character type in step A5 according to the determination of the passage of the predetermined time tout in step A4, the handwritten input coordinate data memory 19c in the RAM 19 is determined. Based on the handwritten BOX position information 16d stored in advance in the ROM 16, it is identified that this handwritten input has been made in the Katakana priority recognition area 13A1R according to the coordinate range of the handwritten character “K” stored in At the same time, the priority character type is determined to be “katakana” and stored in the priority character type memory 19e in the RAM 19, and displayed in the priority character type display area 13C as shown in FIG. 12C (steps A8 → A9). ).
[0077]
Further, based on the coordinate data of the handwritten character “K” stored in the handwritten input coordinate data memory 19c in the RAM 19, all character / symbol pattern data stored in the dictionary data memory 20 are referred to and fixed. A candidate character with a similarity or higher is retrieved, and the retrieved candidate character is associated with the similarity and character type and stored in the candidate character similarity data memory 19d in the RAM 19 (step A10).
[0078]
Then, the score value “30” stored in the score value memory 19f is added to the similarity of the candidate character of the character type matching the priority character type “Katakana” stored in the priority character type memory 19e, and the candidate The candidate characters in the character similarity data memory 19d are sorted again in descending order of their similarity and displayed in a list in the candidate character display area 13B2 on the liquid crystal display unit 13 (steps A11 and A12).
[0079]
In this case, as candidate characters for the handwritten input character “ke”, for example, “ke” “month” “ku” “ri” are searched, but among the candidate characters, stored in the priority character type memory 19e. The added score “30” is added to the similarity of “K”, “K”, “K” of Katakana, which is the character type that matches the priority character type “Katakana”, and “K” of Katakana is the first candidate. The character, katakana “k” is displayed as a second candidate character, katakana “ku” is displayed as a third candidate character, and hiragana “ri” is displayed as a fourth candidate character in the candidate character display area 13B2. The second candidate character “month” is selected in the input character display area 13B1 by selecting the second candidate character “Ka”.
[0080]
Thus, when the handwritten character “K” is recognized and displayed as “Ka” of katakana, the continuous character input time interval t counted by the continuous character input timer 19g is once reset and restarted ( Step A13).
[0081]
Thereafter, after a predetermined time tout stored in the continuous character input time-out memory 19h has elapsed, as shown in FIG. 12D, the kanji “hill” is displayed in the kanji priority recognition area 13A1 at the left end of the touch panel 15. When handwritten input is performed, the previous preferred character type “Katakana” stored in the preferred character type memory 19e is erased, and the same preferred character type “Katakana” displayed in the preferred character type display area 13C is erased ( Steps A2 to A5).
[0082]
Then, the xy coordinate data corresponding to the handwritten trajectory repeatedly detected at regular time intervals by the tablet 14 accompanying the handwritten input of the handwritten character “hill” is sequentially fetched, and the handwritten input coordinate data memory 19 c in the RAM 19 is loaded. Stored (steps A6, A7 → A6).
[0083]
Here, if it is determined that the priority character type is not stored by the deletion process of the priority character type in step A5 according to the determination of the passage of the predetermined time tout in step A4, the handwritten input coordinate data memory 19c in the RAM 19 is determined. Based on the handwritten BOX position information 16d stored in advance in the ROM 16, it is identified that this handwritten input has been made in the Chinese character priority recognition area 13A1 according to the coordinate range of the handwritten character “hill” stored in FIG. At the same time, the priority character type is determined to be “Kanji” and stored in the priority character type memory 19e in the RAM 19 and displayed in the priority character type display area 13C as shown in FIG. 12E (steps A8 → A9). ).
[0084]
Further, based on the coordinate data of the handwritten character “hill” stored in the handwritten input coordinate data memory 19c in the RAM 19, all character / symbol pattern data stored in the dictionary data memory 20 are referred to and fixed. A candidate character with a similarity or higher is retrieved, and the retrieved candidate character is associated with the similarity and character type and stored in the candidate character similarity data memory 19d in the RAM 19 (step A10).
[0085]
Then, the score value “30” stored in the score value memory 19f is added to the similarity of the candidate character of the character type matching the priority character type “Kanji” stored in the priority character type memory 19e, and the candidate The candidate characters in the character similarity data memory 19d are sorted again in descending order of their similarity and displayed in a list in the candidate character display area 13B2 on the liquid crystal display unit 13 (steps A11 and A12).
[0086]
In this case, as a candidate character for the handwritten input character “hill”, for example, only “hill” is searched, but “Kanji”, which is the character type that matches the priority character type “Kanji” stored in the priority character type memory 19e, is searched. The score “30” is added to the similarity of “hill”, and the “hill” of the kanji is displayed as a list in the candidate character display area 13B2 as the first candidate character, and the input character display area 13B1. Is displayed.
[0087]
Thus, when the handwritten character “hill” is recognized and displayed, the continuous character input time interval t counted by the continuous character input timer 19g is once reset and restarted (step A13).
[0088]
Thereafter, the same character recognition processing in steps A2 to A13 is repeatedly performed, so that the continuous character input time interval t counted by the continuous character input timer memory 19g is stored in the continuous character input timeout memory 19h. Before the predetermined time tout is exceeded, as shown in FIG. 13 (A), in the kanji priority recognition area 13A2 which is right next to the leftmost kanji priority recognition area 13A1 where the handwritten character “hill” is input by handwriting. When handwritten input is “dan”, as shown in FIG. 13B, the priority character type “Kanji” up to the previous time is stored and displayed as it is, and the priority character type is “Kanji”. The first candidate character “group”, the second candidate character “same”, the third candidate character “country”... Each is displayed in the candidate character display area 13B2, the first candidate character "Dan" is displayed in the input character display area 13B1.
[0089]
Then, the continuous character input time interval t in the continuous character input timer memory 19g is reset and started again.
Thereafter, the same character recognition processing in steps A2 to A13 is repeatedly performed, so that the continuous character input time interval t counted by the continuous character input timer memory 19g is stored in the continuous character input timeout memory 19h. In the state where the predetermined time tout is exceeded, the handwritten characters “hill” and “group” are input by handwriting in the kanji priority recognition area 13A3 which is further to the right of the kanji priority recognition areas 13A1 and 13A2. When handwriting is input, the priority character type “Kanji” up to the previous time is once erased, as shown in FIGS.
[0090]
Then, as shown in FIG. 13E, the priority character type is determined to be “kanji” according to the kanji priority recognition area 13A3 to which the handwritten character “ground” is input, and is stored and displayed. A first candidate character “ground” and a second candidate character “pond” sorted by adding a numerical value to each similarity of the characters are displayed in the candidate character display area 13B2, and the first candidate character “ground” is displayed. It is displayed in the input character display area 13B1.
[0091]
Then, the continuous character input time interval t in the continuous character input timer memory 19g is reset and started again.
In this way, even if the priority recognition areas 13A1 to 13A4 and 13A1L to 13A2R of different character types are set to overlap, the size of each area can be made different so that handwritten characters can be recognized in the priority recognition areas of any character type. The character type of the handwritten input character can be changed by determining whether or not the input has been performed, measuring the time interval t of continuous character input, and determining whether or not the time interval t has exceeded a predetermined time tout. And whether to change the priority recognition character type set in the input area at that time, or perform the priority recognition processing, or perform the priority recognition processing with the previous priority recognition character type. Therefore, even when inputting continuous handwritten characters with different character types, natural handwritten characters can always be input from left to right. , Each of the handwritten characters is added as a priority recognition of the corresponding character type, and even if the handwritten character has a confusing character between different character types, the correct character is recognized and displayed as the first candidate character. .
[0092]
FIG. 14 is a diagram showing a series of handwritten character input display operations by continuous character input on the touch panel 15 and a change display operation of the priority recognized character type according to the character recognition processing of the second embodiment of the handwritten character recognition apparatus.
[0093]
That is, when continuous handwritten character input with different character types is performed, each time the character type is changed, the priority recognition area of the corresponding character type is found for the first character, and handwritten character input is sequentially performed from that area to the right area. By doing so, each character input by handwriting is preferentially recognized by the character type of the character, and correct character recognition can be performed by a natural character input procedure.
[0094]
In the handwritten character recognition apparatus according to the second embodiment, all of the four handwritten character input areas divided into four sections on the touch panel 15 are set as the kanji priority recognition areas 13A1 to 13A4, respectively, and the kanji character priority recognition areas 13A1 to 13A4 are set. Further, the left and right half of the kanji priority recognition area 13A1 is further divided into the left half area and the right half area as the hiragana priority recognition area 13A1L, and the right half area as the katakana priority recognition area 13A1R. By setting the left half area of the priority recognition area 13A2 to the alphabetic priority recognition area 13A2L and the right half area to the number / symbol priority recognition area 13A2R, even if priority recognition areas of different character types overlap. By changing the size of each overlapping area, However, in the touch panel 15, each of the handwritten character input areas 13A1 to 13A4 divided into four sections partially overlaps the input range for each different character type. It is good also as a structure which discriminate | determines which character type is set as the object of priority recognition by judging in which range the input of the handwritten character was performed, setting by shifting in a state.
[0095]
(Third embodiment)
FIG. 15 is a diagram showing a setting state of a handwritten character input range for each character type in each of the handwritten character input areas 13A1 to 13A4 divided into four sections on the touch panel 15 according to the third embodiment of the handwritten character recognition apparatus.
[0096]
In the third embodiment of the handwritten character input range setting in this handwritten character recognition apparatus, there are nine divided unit areas divided into three vertically and horizontally in each of the handwritten character input areas 13A1 to 13A4 divided into four sections on the touch panel 15. As shown in FIG. 15 (A), the entire nine divisions (1 to 9) are set to the kanji priority recognition range, and the upper left four divisions (1, 2, 4, 5) as shown in FIG. 15 (B). Hiragana priority recognition range, as shown in FIG. 15C, the upper right four sections (2, 3, 5, 6) are Katakana priority recognition range, and as shown in FIG. 15D, the lower left four sections ( 4, 5, 7, 8) are set as the alphabetic priority recognition range, and as shown in FIG. 15E, the lower right four sections (5, 6, 8, 9) are set as the number / symbol priority recognition range.
[0097]
FIG. 16 is a diagram showing a series of handwritten character input display operations by continuous character input on the touch panel 15 and a change state of the priority recognized character type in accordance with the character recognition process of the third embodiment of the handwritten character recognition apparatus.
[0098]
That is, when performing continuous handwritten character input in which different character types are mixed, as shown in FIGS. 16A to 16C, for example, the hiragana handwritten characters are located at the upper left of each handwritten character input area 13A1 to 13A4. Input according to the range of 4 divisions (1, 2, 4, 5), and the katakana handwritten characters are those of the 4 divisions (2, 3, 5, 6) in the upper right of each handwritten character input area 13A1 to 13A4. Input according to the range, handwritten characters in Chinese characters are input corresponding to the range of all nine categories (1-9) of each handwritten character input area 13A1-13A4, and handwritten characters in English are handwritten The character input areas 13A1 to 13A4 are input in correspondence with the lower left four categories (4, 5, 7, and 8), and the numbers and symbols of the handwritten characters are lower right of the handwritten character input areas 13A1 to 13A4. By inputting in correspondence with the range of 4 divisions (5, 6, 8, 9), not only can continuous input of handwritten characters be performed in a natural input procedure from left to right, but also each handwriting. Correct recognition of characters can be performed by preferentially recognizing all characters with the corresponding character types.
[0099]
In the handwritten character recognition device of each of the first to third embodiments, the touch panel 15 in which the transparent tablet 14 is superimposed on the handwritten character display area 13A has been described as performing input and display of handwritten characters. For example, a mark such as a mouse pointer displayed in the handwritten character display area 13A is moved by operating the handwriting input unit 12a such as a mouse. Arbitrary characters may be input by handwriting and displayed by the locus.
[0100]
Further, the method described in the above embodiments, that is, the character recognition processing according to the first embodiment of the handwritten character recognition apparatus shown in the flowchart of FIG. 5, and the second embodiment of the handwritten character recognition apparatus shown in the flowchart of FIG. Each method such as character recognition processing according to the form includes a memory card (ROM card, RAM card, etc.), a magnetic disk (floppy disk, hard disk, etc.), an optical disk (CD-ROM, DVD) as programs that can be executed by a computer. Etc.) and can be stored and distributed in an external storage medium 18 such as a semiconductor memory. Then, the computer reads the program recorded in the external storage medium 18 by the external storage reading unit 17, and the operation is controlled by the read program, so that the priority recognition for each character type described in the above embodiments is performed. It is possible to realize a handwritten character recognition display function and execute the same processing by the above-described method.
[0101]
【The invention's effect】
As described above, according to the first handwritten character recognition apparatus of the present invention. If handwriting is input continuously within a predetermined time, the character type recognized last time will be recognized regardless of the character type of the current input area. Can be input continuously with a simple input procedure. In this way, characters of the same character type are continuously handwritten in the input area and displayed as candidates. It becomes like this.
[0103]
Therefore, according to the present invention, it is possible to perform natural character input from left to right without fixing the input area for each character type, and it is possible to correctly recognize confusing characters existing between different character types. become.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a configuration of an electronic circuit of a handwritten character recognition apparatus according to an embodiment of the present invention.
FIG. 2 is a diagram showing a configuration of a display area in a liquid crystal display unit of the handwritten character recognition device.
FIG. 3 is a diagram showing a data storage state in a handwritten input coordinate data memory provided in a RAM of the handwritten character recognition apparatus;
FIG. 4 is a diagram showing a data storage state in a candidate character similarity data memory provided in a RAM of the handwritten character recognition apparatus.
FIG. 5 is a flowchart showing character recognition processing in the handwritten character recognition apparatus;
FIG. 6 is a view showing a display operation for handwritten character recognition accompanying a character recognition process of the handwritten character recognition apparatus;
FIG. 7 shows a display operation of the handwritten character when an alphanumeric “3” confusing hiragana “ro” accompanying handwriting recognition processing of the handwritten character recognition device is input in an alphanumeric priority recognition area. Figure.
FIG. 8 shows each of the candidate character similarity data memories in the case where the alphanumeric character “3” confused with the character recognition process of the handwritten character recognition device is handwritten in the alphanumeric character priority recognition area. The figure which shows the similarity memory | storage state of a candidate character.
FIG. 9 is a diagram showing a configuration of a handwritten character display area when a priority recognition area for each character type on the touch panel is set in an overlapping manner according to the second embodiment of the handwritten character recognition apparatus.
FIG. 10 is a flowchart showing a character recognition process of the second embodiment in the handwritten character recognition apparatus.
FIG. 11 is a view showing a display operation (part 1) of handwritten character recognition accompanying character recognition processing of the second embodiment of the handwritten character recognition apparatus;
FIG. 12 is a diagram showing a display operation (part 2) of handwritten character recognition accompanying character recognition processing of the second embodiment of the handwritten character recognition apparatus;
FIG. 13 is a diagram showing a display operation (part 3) of handwritten character recognition accompanying character recognition processing of the second embodiment of the handwritten character recognition apparatus;
FIG. 14 is a diagram showing a series of handwritten character input display operations and a priority recognized character type change display operation by continuous character input on a touch panel accompanying character recognition processing of the second embodiment of the handwritten character recognition apparatus;
FIG. 15 is a diagram illustrating a setting state of a handwritten character input range for each character type in each of the four handwritten character input areas according to the touch panel according to the third embodiment of the touch panel.
FIG. 16 is a diagram showing a series of handwritten character input display operations by continuous character input on a touch panel and a change state of a priority recognized character type in accordance with a character recognition process of the third embodiment of the handwritten character recognition apparatus;
[Explanation of symbols]
11: Control unit (CPU),
12 ... Key input part,
12a ... handwriting input part,
13 ... Liquid crystal display,
13A ... handwritten character display area,
13A1 ... Kanji priority recognition area,
13A2 ... Hiragana priority recognition area,
13A3 ... Katakana priority recognition area,
13A4 ... Alphanumeric priority recognition area,
13B ... Input / candidate character display area,
13B1 ... Input character display area,
13B2 ... candidate character display area,
13C ... priority character type display area,
14 ... Transparent tablet,
15 ... Touch panel,
16 ... ROM,
16a ... system program,
16b ... various application programs,
16c ... character recognition processing program,
16d: Handwritten BOX position information,
17 ... external memory reading unit,
18 ... an external storage medium,
19 ... RAM,
19a ... system data memory,
19b Application data memory,
19c ... handwritten input coordinate data memory,
19d ... candidate character similarity data memory,
19e ... priority character type memory,
19f ... Additional numerical value memory,
19g ... continuous character input timer memory (t),
19h ... continuous character input timeout memory (tout),
19i Work memory,
20 ... dictionary data memory,
21 ... Timer,
k ... cursor.

Claims (4)

A handwritten character input means having a plurality of handwritten character input areas, and character types that are preferentially recognized for each area;
After the recognition of the input of the previous handwritten character, when the current handwritten character is input to the handwritten character input means within a predetermined time, the character type recognized last time is also preferentially recognized this time. Continuous input recognition character type setting means ;
An input area recognition character type setting unit for preferentially recognizing the character type set in the area where the handwritten character is input when the current handwritten character is input to the handwritten character input unit at a time other than within the time;
The handwritten character recognition apparatus characterized by comprising. In the handwritten character input means, a priority recognition range of the character type that is different in position according to the character type of the handwritten character in each area is set in advance .
The handwritten character recognition apparatus according to claim 1, wherein the priority recognition ranges partially overlap each other. A handwritten character recognition method of a handwritten character recognition device having a plurality of handwritten character input areas and pre-set character types to be recognized with priority for each area,
A step of setting a continuous input recognition character type for preferentially recognizing the last recognized character type when the current handwritten character is input within a predetermined time after recognition of the previous input of the handwritten character. When,
An input area recognition character type setting step for preferentially recognizing the character type set in the area where the handwritten character is input when the current handwritten character is input at a time other than within the time;
A handwritten character recognition method characterized by comprising: The input area of a plurality of handwritten characters, a storage medium storing handwriting recognition program for controlling a computer handwriting recognition device character type recognizing priority is set in advance in each area,
The computer,
Continuous input recognition character type setting means for recognizing and searching for the previously recognized character type this time when the current handwritten character is input within a predetermined time after recognition for the previous input of the handwritten character ,
Input area recognition character type setting means for preferentially recognizing the character type set in the area where the handwritten character is input when the current handwritten character is input at a time other than within the time period,
A storage medium storing a handwritten character recognition program for functioning as a computer.

Images