Character recognition device, character recognition method and program

Abstract

To efficiently utilize handwritten information so that a desired character can be selected from a wide range of candidates.SOLUTION: A character recognition apparatus includes: a touch-panel display unit 17 which receives a character image based on a handwritten input operation; and a control unit 21 which executes recognition processing on the received character image to derive a first character from the character image, and causes the touch-panel display unit 17 to display the derived first character as a candidate of character desired by a user. The touch-panel display unit 17 can receive a designation operation indicating that the first character is not the character desired by the user but a character similar to the character desired by the user. The control unit 21 derives, when receiving the designation operation, a second character having a predetermined relationship with the first character, and causes the touch-panel display unit 17 to display the derived second character as an additional candidate of the character desired by the user.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to a character recognition device, a character recognition method, and a program. [Background technology]

[0002] A technology has been proposed that, for a series of character strings that have been handwritten and recognized, allows the recognition candidate characters to be displayed in sequence for each character to be corrected, making it easy to correct the character to a desired recognition candidate character without the need for a selection operation (for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2008-299431 A Summary of the Invention [Problem to be solved by the invention]

[0004] In the case where the desired character is not found among a plurality of characters that are candidates for correction, including the techniques described in the patent documents, it is necessary to carry out character recognition processing again on the handwritten character.

[0005] The present invention has been made in view of the above-mentioned circumstances, and an object of the present invention is to provide a method for converting handwritten input data into a text. Character image of the character By making effective use of Candidates similar to the recognized characters are derived from the handwritten input, and from the derived character candidates A character recognition device, a character recognition method, and a program capable of accepting the selection of a desired character. [Means for solving the problem]

[0006] One aspect of the present invention includes an input unit that receives a character image of a first character input by handwriting, and a recognition process that executes a recognition process on the character image received by the input unit; Character candidates are sorted in order of similarity to the character image. A character derivation unit that derives a character Character suggestionsa display control unit that displays a character as a candidate for a character desired by a user, The display control unit selects a character that the user desires from among the character candidates displayed by the display control unit. The character deriving unit is capable of receiving a designation operation, and when the designation operation is received by the input unit, the character deriving unit generates stroke data of the character image received by the input unit. To Corrected stroke data of the stroke data of the character image of the first character is obtained from the character shape of the candidate selected as the most similar, and a recognition process is performed on the character image based on the corrected stroke data to derive a second character, and the display control unit displays the second character derived by the character derivation unit as an additional candidate for the character desired by the user. Effect of the Invention

[0007] According to the present invention, handwritten input Character image of the character By making effective use of Candidates similar to the recognized characters are derived from the handwritten input, and from the derived character candidates A selection of the desired character is then available for acceptance. [Brief description of the drawings]

[0008] [Figure 1] 1 is a diagram showing the external configuration of an electronic dictionary according to an embodiment of the present invention; [Diagram 2] FIG. 2 is a block diagram showing the functional configuration of an electronic circuit of the electronic dictionary according to the embodiment. [Diagram 3] 10 is a flowchart showing processing for handwritten input characters according to the embodiment. [Figure 4] 4 is a flowchart of a subroutine showing detailed processing contents of the stroke data correction processing in FIG. 3 according to the embodiment; [Diagram 5] 6A and 6B are diagrams illustrating examples of screens showing handwritten character input and its recognition result according to the embodiment. [Figure 6] 6A to 6C are diagrams showing a process of deriving a plurality of correction stroke data according to the embodiment; [Figure 7] 4A to 4C are diagrams showing screen transitions of the touch panel display unit according to the embodiment. [Figure 8] 4A to 4C are diagrams showing screen transitions of the touch panel display unit according to the embodiment. [Figure 9] 13A and 13B are diagrams illustrating examples of some of the biased tables stored in the various table storage areas according to the embodiment. [Figure 10] 11A to 11C are views for explaining a process of deriving radical-based kanji candidates from stroke data input by a user according to the embodiment; [Figure 11] 11A and 11B are diagrams showing an example in which stroke data input by a user is replaced with recent components of a character shape according to the embodiment; DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] Hereinafter, an embodiment in which the present invention is applied to an electronic dictionary will be described with reference to the drawings. [composition] 1 is a front view showing the external configuration of an electronic dictionary 10 according to the embodiment. This embodiment can be configured not only as the electronic dictionary 10 described below, but also as a tablet-type PC (Personal Computer), smartphone, electronic book, portable game machine, server on a communication network, etc., equipped with a dictionary function.

[0010] The electronic dictionary 10 is configured with a foldable case in which the main body case 11 and the cover case 12 can be opened / closed via a hinge section 13. On the surface of the main body case 11 when the foldable case is opened, there are provided a key input section 14 (hardware keys) including a [HOME] key 14a, a function designation key 14b, a character input key 14c, a [ENTER] key 14d, a [BACK] key 14e, a [BOX] key 14f, a cursor movement key 14g, a [SHIFT] key 14h, etc., an audio output section (including a speaker) 15, an audio input section (including a microphone) 16, etc.

[0011] A touch panel display unit 17 is provided on the surface of the cover case 12. The touch panel display unit 17 is an integrated structure of a touch position detection device that detects the position touched by the user of the electronic dictionary 10 with a pen or finger, and a display device, and is configured by laminating a transparent touch panel on a backlit liquid crystal screen. That is, the touch panel display unit 17 functions as an input unit that accepts character images based on input operations in handwritten format.

[0012] The [Home] key 14a of the key input unit 14 is a key for displaying a home screen on the key input unit 14. Although not specifically shown, a plurality of icons registered according to the initial settings of the electronic dictionary 10 or user operations are displayed. Each icon is a drawing or symbol for realizing a function corresponding to the notation of the icon. For example, it includes an application icon for directly starting a function (application) that uses dictionary content or learning content, and a group icon for displaying a list screen of icons of a plurality of functions belonging to one category.

[0013] The function designation keys 14b of the key input unit 14 are keys for directly designating dictionary content etc. displayed on each key. The function designation keys 14b include a [Multiple Dictionaries] key that does not specify a category of dictionary content, keys for designating the same category (such as the [Japanese] key, the [Archaeological] key, the [Chinese-Japanese] key, and the [English-English-Japanese] key), a [Content List] key that displays a list of dictionary contents, and a [Study Notebook] key, which is one category of tools.

[0014] Furthermore, when a key of the key input unit 14 is operated after the [Shift] key 14h is operated, the key functions as a key enclosed in a frame on the key top, not as a key without a frame. For example, when the [Delete] key is operated after the [Shift] key 14h is operated, the key functions as a [Setting] key, not as a [Delete] key.

[0015] FIG. 1 shows a state in which [Japanese Dictionary] is selected, and in the search string input section 41 located at the top of the touch panel display section 17, the kana input tag is selected out of the kana input tag (Aiu) and the alphabet input tag (ABC).

[0016] The character string [Nikko] has already been input into the search character string input section 41. A candidate display section 42 below the search character string input section 41 shows a state in which, for example, three entry words corresponding to the character string [Nikko] are displayed as a list. Furthermore, below the candidate display section 42, a handwritten character input section 43 for two characters is displayed, showing a blank state in which the section is waiting for the handwritten input of characters following the character string [Nikko].

[0017] FIG. 2 is a block diagram showing the functional configuration of the electronic circuit of the electronic dictionary 10. As shown in FIG. The electronic circuit of the electronic dictionary 10 includes a control unit (CPU: Central Processing Unit) 21, which is a computer. The control unit 21 controls the operation of each part of the circuit according to a control program prestored in a storage unit 22 such as a flash ROM. There may be one processor such as a CPU or two or more processors.

[0018] At least one of the control program and the dictionary data described later may be read from an external recording medium 23 such as a memory card by the recording medium reading unit 24 and stored in the memory unit 22, or may be downloaded from an external web server on a network (not shown) via the communication unit 25 and stored in the memory unit 22.

[0019] The control unit 21 is connected to a memory unit 22, a recording medium reading unit 24, and a communication unit 25 via a data bus and a control bus, as well as to a key input unit 14, an audio output unit 15, an audio input unit 16, and a touch panel display unit 17.

[0020] The memory unit 22 stores a system program that controls the overall operation of the electronic dictionary 10, a communication program for communicating with external devices via the communication unit 25, as well as memory areas for storing programs and data for executing various functions by the electronic dictionary 10, such as a search processing program memory area 22a, a dictionary data memory area 22b, a search history memory area 22c, and a working data memory area 22d.

[0021] The search processing program memory area 22a is an area for storing a control program for searching for and displaying a user-requested headword and its corresponding translation, meaning, usage example, explanation, and other explanatory information based on various dictionary contents (English-English-Japanese dictionary / English-Japanese dictionary / Japanese-English dictionary / English-English dictionary / Japanese dictionary / encyclopedia / ...) stored as dictionary data in the dictionary data memory area 22b, as well as various table memory areas 22a1 used when executing a search in response to an input other than the character input of the headword.

[0022] The various table storage area 22a1 stores a plurality of tables including a radical type table that predetermines how radicals, which are components of each character data of a Chinese character, can be divided.

[0023] When an entry word and its explanatory information are displayed as search results in accordance with a search in response to a user operation, the search history storage area 22c stores the entry word that is the subject of the search as a search history together with information on the number of times the entry word has been searched.

[0024] In the working data memory area 22d, various data input in response to user operations and data acquired or generated by the control unit 21 are temporarily stored (held) as needed in accordance with control of the operation of each part of the circuit in accordance with the control program by the control unit 21.

[0025] The communication unit 25 wirelessly connects to external devices such as similar electronic dictionaries in the vicinity based on the BLE (Bluetooth (registered trademark) Low Energy) standard, which is one of the wireless PAN (Personal Area Network) technologies, and also wirelessly connects to web servers and the like on an external network (not shown) via a wireless router or the like to download control programs and the like.

[0026] [Operation] Next, the operation of this embodiment will be described. In this embodiment, a correction process will be described in the case where the recognition result of the handwritten input character image when the Japanese dictionary function is selected is not the kanji desired by the user.

[0027] Here, we will explain a case where, for example, the character image of the character string [sunlight] is input by handwriting as a search string and is recognized as desired, and then the character image of the character [bath] is similarly input by handwriting, but the desired character [bath] is not found among the character candidates in the recognition result of the input character image.

[0028] FIG. 5(A) shows the screen when the user of the electronic dictionary 10 further inputs the character image of the character [bath] in handwriting input section 43 using stylus pen P, from the display state of touch panel display section 17 shown in FIG.

[0029] As shown in the figure, handwritten character input unit 43 has an input area for two characters. After handwritten input is performed on the left side, for example, a stylus pen P is used on the right side to instruct execution of recognition processing for the input characters.

[0030] In response to an execution instruction for recognition processing with the stylus pen P, the control unit 21 executes character recognition processing and displays a list of character candidates, which are the recognition results, in the order of being evaluated as having a high similarity. That is, the control unit 21 executes recognition processing on the character image received by the input unit, functions as a character derivation unit that derives a first character from the character image, and a display control unit that displays the first character derived by the character derivation unit as a candidate for the character desired by the user.

[0031] FIG. 5(B) shows a state in which Chinese character candidates evaluated as having a high similarity are listed and displayed on the candidate character display unit 44. Since the desired Chinese character [bath] is not included in the candidate character display unit 44, the user needs to perform an operation for correcting the handwritten input character.

[0032] FIG. 3 is a flowchart showing the processing contents for the handwritten input character executed by the control unit 21.

[0033] At the beginning of the processing, the control unit 21 acquires information on candidate characters corresponding to the recognition result (step S101), and as shown in FIG. 5(B), lists and displays those candidate characters on the candidate character display unit 44.

[0034] From the state in which these Chinese character candidates are listed and displayed, the control unit 21 repeatedly determines whether one of the candidates considered to be the closest has been long-pressed (step S103), whether one candidate has been selected (step S107), and whether a cancel operation has been performed on the list display (step S109), and waits for any of these inputs to be made.

[0035] In step S107, when it is determined that a cancel operation has been performed on the list display (YES in step S109), the control unit 21 ends the processing of FIG. 3 as described above.

[0036] FIG. 7(A) shows a state in which the user is long-pressing the Chinese character candidate [Qia] in the handwritten character input unit 43 with the stylus pen P on the screen of the touch panel type display unit 17 shown in FIG. 5(B).

[0037] In step S103, when it is determined that a long-press operation has been performed on one candidate, the control unit 21 executes a stroke data correction process to derive a plurality of corrected stroke data from the original input stroke data input in handwriting and the character shape of the candidate selected as the one with the highest similarity (step S104).

[0038] FIG. 6 is a diagram illustrating the process of deriving a plurality of corrected stroke data. In accordance with the stroke data of the character input in handwriting shown in FIG. 6(A) and the shape of the character [Qia] selected as the closest one judged by the user shown in FIG. 6(B), the control unit 21 derives the corrected stroke data 1 shown in FIG. 6(C) and the corrected stroke data 2 shown in FIG. 6(D).

[0039] FIG. 4 is a flowchart of a subroutine showing the details of the correction process of the stroke data executed by the control unit 21 in step S104. First, the input stroke data of the user is scanned to derive glyph analysis data including an outer shape frame, blank information, and density information (step S201).

[0040] The control unit 21 refers to the radical type table stored in the various table storage areas 22a1 based on the derived glyph analysis data, and derives radical type candidates for the stroke data input by the user (step S202).

[0041] FIG. 9 is a diagram illustrating a part of the radical type table stored in the various table storage areas 22a1. As shown in the figure, the radical compositions of Chinese characters can be classified into those consisting of the radical and side shown in FIG. 9(A), those consisting of the crown and leg shown in FIG. 9(B), those including the winding (Nyou) shown in FIG. 9(C), those including the hanging (Tare) shown in FIG. 9(D), those including the other form (Kamae) shown in FIG. 9(E), those that cannot be classified into other radicals shown in FIG. 9(F), and the like.

[0042] FIG. 10 is a diagram for explaining the process of deriving radical-based Chinese character candidates from the stroke data input by the user. As shown in FIG. 10(A), for the stroke data of the user input shown in FIGS. 5(A) and 6(A), sequential scanning is performed in one direction each in the vertical and horizontal directions.

[0043] FIG. 10(B) shows the results of deriving the outer frame OF, blank information SI, and dense information CI of the stroke data based on the coarseness and fineness of the character image from the scanning results. From these derived results, as shown in FIG. 10(C), it is recognized as being composed of two radicals, the left radical Rix1 and the right radical Rix2 shown in FIG. 9(A), and considering the long-pressed Chinese character candidate, a highly similar radical-based candidate is derived.

[0044] In FIG. 10(C), while the stroke data input by the user is recognized as being composed of the left and right radicals shown in FIG. 9(A), in FIG. 7(A), the Chinese character candidate [Qia] in the handwritten character input section 43 with the stylus pen P by the user is also composed of the left and right radicals shown in FIG. 9(A). Thus, when the radical type judged from the handwritten input character image matches the radical type of the selected character with a closer glyph, the radical-based candidate is limited to one. On the other hand, when the radical type judged from the handwritten input character image does not match the radical type of the selected character with a closer glyph, each is derived as a radical-based candidate. That is, two radical-based candidates are derived. The control unit 21 selects one unselected candidate from the derived radical-based candidates (step S203). If the control unit 21 selects one of the unselected candidates, for example, the left and right radicals shown in FIG. 9(A), as the number of components of the stroke data for the selected radical-based candidate, for example, [2] is derived from [left] and [right] (step S204).

[0045] Next, the control unit 21 selects one of the components of the user input, for example, [left] (step S205). For the selected component [left], the control unit 21 selects Nearest character (stroke data of the handwritten input character shown in FIG. 6(A) and the character selected by the user as being closest to the stroke data shown in FIG. 6(B)) one of the components of [Qia], for example, [Sanshui (Three Waters)] (step S206).

[0046] The control unit 21 further determines whether it is possible to omit processing such as replacement, which will be described later, for the selected component, for example, [sanzui (three waters)] (step S207). This is a measure to reduce the processing steps of unnecessary user input stroke data and replacement of each component of the nearest character, etc., by setting constraints such as not arranging a biased component next to another component.

[0047] In step S207, for example, when the component selected by the user input stroke data is [biased], and the component selected as the component of the nearest character [qia] is [he] on the [side] side, It is determined that the process of replacement etc. can be omitted. (YES in step S207), the control unit 21 returns to the processing from step S206, assuming that the processing for the component selected at that time will be omitted, and proceeds to the next Substitution etc. selection. Components of the most recent character shapes

[0048] Also in step S207, when it is determined that the selected component cannot be omitted (NO in step S207), the control unit 21 replaces the [biased] - side component selected by the user input stroke data with the same component [sanzui (three waters)] of the nearest character [qia] (step S208). Then, the Chinese character candidates based on the replaced component are saved as corrected stroke data (step S209).

[0049] The control unit 21 determines whether all Recent character components selections have been completed, to confirm that there are components of the nearest unselected characters and that not all Recent character components have been selected after all selections are completed (step S210).

[0050] In step S210, when it is determined that not all Recent character components selections have been completed (NO in step S210), the process returns to the processing from step S206, and recent characters processing corresponding to other components is executed.

[0051] In this way, with respect to the components on the side of the stroke data of the user input as well, a process is executed in which the components on the side of the nearest character [洽] are replaced with those on the side of the nearest character [旁] to obtain corrected stroke data.

[0052] In step S210, when all Modern character components the selections are completed and it is determined that there are no components of the nearest character that have not been selected (YES in step S210), the control unit 21 determines whether all the components of the stroke data of the user input have been selected, thereby confirming that there are unselected components and that all the components have not been selected after the selection is completed (step S211).

[0053] In step S211, when it is determined that the selection of all the components of the stroke data of the user input has not been completed and there are unselected components (NO in step S211), the control unit 21 returns to the process from step S205 and executes the same process for the other components of the stroke data of the user input.

[0054] Also, in step S211, when it is determined that the selection of all the components of the stroke data of the user input has been completed and there are no unselected components (YES in step S211), the control unit 21 determines whether the selection of all the radical type candidates has been completed, thereby confirming that there are unselected radical type candidates and that all the candidates have not been selected after the selection is completed (step S212).

[0055] In step S212, when it is determined that the selection of all the radical type candidates has not been completed (NO in step S212), the control unit 21 returns to the process from step S203 and executes the same process based on the unselected radical type candidates.

[0056] Also, in step S212, when it is determined that the selection of all the radical type candidates has been completed (YES in step S212), the control unit 21 ends the subroutine regarding the correction process of the stroke data in FIG. 4 above and returns to the main routine in FIG. 3.

[0057] As described above, when the radical type determined from the handwritten input character image matches the radical type of the character selected as the one with a closer glyph, the radical type candidate is limited to one.

[0058] On the other hand, when the radical type determined from the handwritten input character image does not match the radical type of the character selected as the one with a closer glyph, Two radical candidates are selected: one determined from the handwritten character image and the other selected as the closest character shape. will be derived, so steps S203 and below are repeatedly executed.

[0059] FIG. 11 shows, corresponding to the character "[Qia]" long-pressed by the user in the candidate character display section 44 of the touch panel display section 17, the stroke data of the user input in terms of the recent characters constituent elements of the radical type. For the stroke data of the user input, FIG. 11(A) shows the selected Recent character shape kanji candidates [Qia]. The corrected stroke data 2 corresponding to [Sanshui (さんずい)], which is its partial Rix1, is shown in FIG. 11(C), and similarly, the corrected stroke data 1 corresponding to [He], which is its side Rix2, is shown in FIG. 11(B). In FIG. 11, the final number of corrected stroke data is [2], and character recognition processing described later is performed on them.

[0060] In FIG. 3, after executing the correction process of the stroke data in step S104, the handwriting recognition process for a plurality of corrected data is sequentially executed to derive corrected Chinese character candidates as recognition results for the plurality of corrected data (step S105).

[0061] The control unit 21 redisplays the derived corrected Chinese character candidates in a list on the touch panel display section 17 (step S106), and then returns to the process from step S103 to wait for an operation on the displayed Chinese character candidates.

[0062] FIG. 7(B) is a diagram illustrating a state in which, from the state where the Chinese character "[Qia]" is selected by a long-press operation with the stylus pen P in FIG. 7(A), Chinese character candidates are displayed with reference to stroke data input by handwriting through the above-described series of processes. In FIG. 7(B), with the center of the corrected Chinese character candidate display section 45 as the position of the selected Chinese character "[Qia]", two Chinese character candidates "[Ha]" and "[Qia]" corresponding to "[He]" are listed and displayed on the left side 45A thereof. Similarly, two Chinese character candidates "[Yu]" and "[Yan]" are listed and displayed on the right side 45B thereof with the Chinese character "[Qia]" in between. The two Chinese character candidates "[Ha]" and "[Qia]" on the left side 45A are those in which candidates derived by performing the handwriting recognition process in step S105 are displayed with respect to the corrected stroke data 1 shown in FIG. 6(C). On the other hand, the two Chinese character candidates "[Yu]" and "[Yan]" on the right side 45B are those in which candidates derived by performing the handwriting recognition process in step S105 are displayed with respect to the corrected stroke data 2 shown in FIG. 6(D).

[0063] In this case, Chinese characters of the same radical type, for example, "[Ha]" and "[Qia]", "[Yu]" and "[Yan]" may be expressed as being of the same radical type by surrounding them with frames of the same color for display.

[0064] Also, not limited to colors, by performing classification display so that the type of frame line, display position, etc. can be visually distinguished for each of the same radical types, the user can easily understand the result of character recognition.

[0065] When the user of the electronic dictionary 10 finds a desired Chinese character among the Chinese character candidates listed and displayed, the user selects the desired Chinese character by a touch operation.

[0066] As described above with reference to FIG. 3, the control unit 21 repeatedly determines whether one of the candidates considered to be the closest is long-pressed (step S103), whether one candidate is selected (step S107), and whether a cancel operation is performed on the list display (step S109), and waits for any of these inputs to be made.

[0067] Therefore, when a single character desired by the user is selected by a touch operation, the control unit 21 determines in step S107 that one candidate has been selected (YES in step S107), confirms the selected kanji candidate, and inputs it following the character string already input (step S108), thereby completing the processing in FIG. 3.

[0068] 8(A) shows a state where the kanji candidate [bath] on the right side 45B of the corrected kanji candidate display section 45 is touched from the display state of FIG. 7(A). In response to this touch operation, the control section 21 determines the kanji candidate [bath] as the character following the character string [nikko] already input in the search character string input section 41.

[0069] 8(B), in the search character string input unit 41, the character string [nikkoshin] including the erroneously recognized character [shin], which is a provisional input state, is changed to the confirmed result [nikkoyaku] after the correction process. At the same time, in the handwritten character input unit 43, the display content is cleared in preparation for the next handwritten input after the kanji is confirmed.

[0070] In this way, even if the desired kanji is not recognized by handwriting input, by performing an operation other than a normal touch operation, such as long pressing another character that is thought to be closest to the recognition result, the stroke data input by the user is decomposed into each radical of the kanji to obtain corrected stroke data, and new character analysis is performed. Therefore, even if part of the kanji to be handwritten input is input somewhat inaccurately, it is possible to improve the possibility of obtaining the desired kanji by correcting it, which contributes to the efficiency of handwriting input.

[0071] Furthermore, when selecting a dictionary function such as a Japanese dictionary, the system takes into account the characters already entered before and after the handwritten characters in the search string that the user enters as part of the headword, and excludes from the list those characters that are deemed unlikely to be candidates, thereby preventing an unnecessarily large number of candidates from being displayed.

[0072] Furthermore, if the desired kanji character is still not displayed in either of the kanji candidates 45A, 45B in the correction kanji candidate display section 45 on the display screen of the touch panel display section 17 shown in Figures 7(B) and 8(A), the character that is thought to be closest in shape is selected from among them again and pressed and held down, thereby repeatedly executing the stroke data correction process shown in Figure 4.

[0073] Furthermore, if the desired kanji character is not displayed as a candidate even after repeatedly selecting kanji characters that are thought to be similar in shape by a long press operation, it is possible to re-execute the handwritten input itself in the handwritten character input unit 43. When re-executing the handwritten input in this manner, the recognition accuracy can be improved by performing a selection process that reduces the appearance rate as candidates of characters that were previously displayed as candidates but were not selected as the desired character.

[0074] [Effects of the embodiment] As described above in detail, according to this embodiment, it is possible to effectively utilize handwritten input contents and accept the selection of desired characters from a wide range of candidates.

[0075] In particular, in this embodiment, stroke data handwritten by a user is analyzed by referring to the radical type table stored in the various table storage area 22a1 to analyze how the components of each radical can be divided, and the corrected stroke data is derived as radical type candidates. By adopting such a processing process, other image processing such as weighting based on learning in character recognition processing or machine learning data may also be performed, but in any case, the load of circuit processing in the control unit 21 can be reduced and the recognition result can be obtained more quickly.

[0076] Although not described in this embodiment, when the user finally selects a desired character candidate, the selected character and the handwritten input character may be divided according to the radical type, and the correspondence between each divided portion may be updated and stored as learning data in the various table storage area 22a1. In this case, the various table storage area 22a1 functions as a storage unit that stores information indicating the relationship between the divided portions of the handwritten input character and the corresponding portions of the finalized character, and stores the information as information to be used when the control unit 21 newly recognizes characters. When a similar handwriting is included in the character recognition of the handwritten input from the next time onwards, the automatic correction is performed according to the learned content, so that the user's handwriting habits can be learned as the user uses the character more. This can further improve the recognition rate.

[0077] In addition, the present invention is not limited to the above-mentioned embodiment, and various modifications can be made in the implementation stage without departing from the gist of the invention. In addition, each embodiment may be implemented in combination as appropriate as possible, and in that case, the combined effect can be obtained. Furthermore, the above-mentioned embodiment includes inventions at various stages, and various inventions can be extracted by appropriate combinations of the disclosed multiple constituent elements. For example, even if some constituent elements are deleted from all the constituent elements shown in the embodiment, if the problem described in the "Problem to be Solved by the Invention" column can be solved and the effect described in the "Effect of the Invention" column can be obtained, the configuration from which the constituent elements are deleted can be extracted as the invention.

[0078] The invention as originally claimed in the present application is set forth below. [Claim 1] an input unit that accepts character images based on an input operation in handwriting format; a character deriving unit that executes a recognition process on the character image received by the input unit and derives a first character from the character image; a display control unit that displays the first character derived by the character derivation unit as a candidate for a character desired by a user; Equipped with the input unit is capable of accepting a designation operation indicating that the first character is not a character desired by the user but is similar to the character desired by the user; the character deriving unit further derives a second character having a predetermined association with the first character when the specifying operation is accepted by the input unit; The display control unit displays the second character derived by the character derivation unit as an additional candidate for a character desired by the user. A character recognition device comprising: [Claim 2] The character derivation unit further derives the second character by using a correlation with the character image received by the input unit. 2. The character recognition device according to claim 1, [Claim 3] the character deriving unit derives the second character for each portion divided according to density of the character image in accordance with a predetermined correlation with a corresponding portion of the first character; 3. The character recognition device according to claim 1 or 2. [Claim 4] the character derivation unit derives a plurality of the second characters having a predetermined correlation with the first character in correspondence with portions divided according to density of the character image; The display control unit displays the derived second characters in a divided manner corresponding to the portions divided according to the density of the character image. 4. The character recognition device according to claim 1, wherein the character recognition device further comprises a processor. [Claim 5] the character derivation unit, when the character image received by the input unit is input following a previously input character string, derives the second character in accordance with a relevance to the previously input character string. 5. The character recognition device according to claim 1, wherein the character recognition device further comprises: [Claim 6] the input unit is capable of accepting a confirmation operation for the second character displayed as an additional candidate, A storage unit stores information indicating a relationship between a portion divided according to density of the character image and a corresponding portion of the second character for which the confirmation operation has been received, as information used by the character derivation unit when deriving the second character. 6. The character recognition device according to claim 1, wherein the character recognition device further comprises: [Claim 7] A character recognition method for a device including an input unit that accepts a character image based on an input operation in a handwritten format, a character derivation unit that executes a recognition process on the character image accepted by the input unit and derives a first character from the character image, and a display control unit that displays the first character derived by the character derivation unit as a candidate for a character desired by a user, the input unit is capable of accepting a designation operation indicating that the first character is not a character desired by the user but is similar to the character desired by the user; the character deriving unit further derives a second character having a predetermined association with the first character when the specifying operation is accepted by the input unit; The display control unit displays the second character derived by the character derivation unit as an additional candidate for a character desired by the user. A character recognition method comprising: [Claim 8] A program executed by a computer having a device including: an input unit that accepts a character image based on an input operation in a handwritten format; a character derivation unit that executes a recognition process on the character image accepted by the input unit and derives a first character from the character image; and a display control unit that displays the first character derived by the character derivation unit as a candidate for a character desired by a user, the program being executed by the computer: the input unit is capable of accepting a designation operation indicating that the first character is not a character desired by the user but is similar to the character desired by the user; the character deriving unit further derives a second character having a predetermined association with the first character when the specifying operation is accepted by the input unit; The display control unit displays the second character derived by the character derivation unit as an additional candidate for a character desired by the user. The program is characterized by causing the program to function as follows. [Explanation of symbols]

[0079] 10. Electronic dictionary 11…Main body case 12...Cover case 13...Hinge part 14...Key input section 14a…[Home] key 14b...Function designation key 14c...Character input keys 14d…[Enter] key 14e…[Back] key 14f…[BOX] key 14g...Cursor movement keys 14h...[Shift] key 15...Audio output section 16…Audio input section 17…Touch panel display 21...Control unit (CPU) 22...Storage section 22a...Search processing program storage area 22a1…Various table storage areas 22b…Dictionary data storage area 22c…Search history storage area 22d…Work data storage area 23...External recording medium 24...Recording medium reading unit 25…Communications Department 41...Search string input section 42...Candidate display area 43…Handwritten character input section 44...Candidate character display area 45…Corrected kanji candidate display area 45A, 45B…Kanji candidates P…Stylus pen

Claims

1. an input unit that accepts a character image of a first character input by handwriting; a character derivation unit that executes a recognition process on the character image received by the input unit and derives character candidates in the order of the degree of similarity evaluated to the character image; a display control unit that displays the character candidates derived by the character derivation unit as character candidates desired by a user; Equipped with The input unit is capable of receiving a designation operation for designating derivation of a character desired by the user from character candidates displayed by the display control unit, the character derivation unit, when receiving the designation operation by the input unit, acquires corrected stroke data of the stroke data of the character image of the first character from a character shape of a candidate selected as a candidate having the highest similarity to the stroke data of the character image received by the input unit, and derives a second character by performing a recognition process on the character image based on the corrected stroke data; The display control unit displays the second character derived by the character derivation unit as an additional candidate for a character desired by the user. A character recognition device comprising:

2. the character image of the first character is an image of a first Chinese character having a radical composition; 2. The character recognition device according to claim 1, wherein, when the designation operation is accepted by the input unit, the character derivation unit determines a radical type of the first Chinese character to be recognized from the image of the first Chinese character depending on density, and acquires stroke data of a character image corresponding to a radical of the determined radical type as corrected stroke data of the stroke data of the image of the first Chinese character.

3. the display control unit displays the derived second Chinese characters, which are the plurality of second characters, in a manner corresponding to the determined radicals; 3. The character recognition apparatus according to claim 2.

4. The character recognition device according to any one of claims 1 to 3, characterized in that, when the first character accepted by the input unit is one of a string of characters consisting of a plurality of characters, the character derivation unit excludes from the plurality of second characters to be added candidates, second characters that are considered unlikely to be candidates to be included in the string, based on a relationship with other strings of the string other than the first character.

5. A character recognition method for a device including an input unit and a display unit that accepts a character image of a first character input by handwriting, a character derivation unit that executes a recognition process on the character image accepted by the input unit and derives character candidates in an order of evaluation of high similarity to the character image, and a display unit that displays the character candidates derived by the character derivation unit as character candidates desired by a user, receiving, by the input unit, a designation operation for designating derivation of a character desired by the user from character candidates displayed on the display unit; acquiring corrected stroke data of the stroke data of the character image of the first character from a character shape of a candidate selected as a candidate having the highest similarity to the stroke data of the character image accepted by the input unit when the input unit accepts the specifying operation, and executing a recognition process on the character image based on the corrected stroke data to derive a second character; displaying the derived second character on the display unit as an additional candidate for a character desired by the user; A character recognition method comprising:

6. A computer includes an input unit and a display unit that accepts a character image of a first character input by handwriting, a character derivation unit that executes a recognition process on the character image accepted by the input unit, and derives character candidates in an order of evaluation of high similarity to the character image, and a device that displays the character candidates derived by the character derivation unit as character candidates desired by a user on the display unit, receiving, via the input unit, a designation operation for designating derivation of a character desired by the user from character candidates displayed on the display unit; acquiring corrected stroke data of the stroke data of the character image of the first character from a character shape of a candidate selected as a candidate having the highest similarity to the stroke data of the character image accepted by the input unit when the input unit accepts the specifying operation, and executing a recognition process on the character image based on the corrected stroke data to derive a second character; displaying the derived second character on the display unit as an additional candidate for a character desired by the user; A program to achieve this.

Images