CN1551038A - Character recognizing method - Google Patents
Character recognizing method Download PDFInfo
- Publication number
- CN1551038A CN1551038A CNA2004100348162A CN200410034816A CN1551038A CN 1551038 A CN1551038 A CN 1551038A CN A2004100348162 A CNA2004100348162 A CN A2004100348162A CN 200410034816 A CN200410034816 A CN 200410034816A CN 1551038 A CN1551038 A CN 1551038A
- Authority
- CN
- China
- Prior art keywords
- character
- combinations
- mentioned
- character recognition
- directions
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Character Input (AREA)
- Character Discrimination (AREA)
Abstract
To prevent a recognition ratio from lowering caused by the erroneous decision of a merge direction. When it is judged that a merge direction should be changed, recognition processing is performed in a merge direction different from the original merge direction, and the result is replaced. Then, the recognition processing is performed again so that character recognition can be performed under an ideal condition with respect to the merge direction.
Description
Technical field
The present invention relates to the character recognition method.
Background technology
With regard to the input picture that comprises character zone, when carrying out the character recognition processing, always be character zone to be extracted out, judge the combinations of directions of the character of character zone again, carry out character recognition according to the combinations of directions of judging and handle by input picture being carried out zone identification.So-called herein combinations of directions refers to the record direction of character string.Such as, there are vertically a plurality of combinations of directions such as combination and horizontal combination in Japanese and Chinese etc.
Method as the combinations of directions of judging the character string that in character zone, exists, the method that the characteristics of image of with good grounds histogram etc. is judged (such as, No. 3135290, Jap.P.) and detect the expression character the high contiguous pixels group of possibility arrangement method (such as, Jap.P. open communique spy open flat 04-090083 number).
, in existing combinations of directions decision method, character area is more little, just easy erroneous judgement the by accident more.
Judging that " 03 " this character string is the occasion of writing across the page, the direction of reading is from left to right.So recognition result is " 0 ", " 3 ".On the other hand, be the vertical occasion of writing in judgement, because row is to arrange from right to left, recognition result is " 3 ", " 0 ", is backward as character string.Therefore, even the identification of single character is correct, if this combinations of directions result of determination is wrong, character identification result as a whole will be made mistakes.
In No. 3135290, Jap.P. in the method for record, be to be prerequisite there to be to a certain degree number of characters.Therefore, judge that the character zone of object is more little,, can become the reason of erroneous judgement because the histogram that combinations of directions causes becomes unstable.
Open communique spy opens in the method for record in flat 04-090083 number at Jap.P., in the character zone that the character zone that comprises the separating character of being made up of independent radicals by which characters are arranged in traditional Chinese dictionaries such as " い " of Japanese, " changes " and double byte character and half-angle character (" l " of English digital, " a " etc.) mix, the zone is more little, just easy erroneous judgement the by accident more.
Summary of the invention
The present invention addresses the above problem and the invention that proposes, and its purpose is to reduce the low situation of discrimination that causes owing to the erroneous judgement to combinations of directions.
The formation of character recognition method of the present invention comprises: the image of input is carried out zone identification and step that character zone is extracted; Judge the combinations of directions determination step of the combinations of directions of the above-mentioned character zone that extracts; And the combinations thereof direction that obtains according to the utilizing above-mentioned judgement character recognition step that above-mentioned character zone carried out character recognition; It is characterized in that the step that comprises is in this character recognition method: the whether consistent condition criterion step of first condition of judging the result utilize the character recognition that above-mentioned character recognition step obtains and regulation; And when consistent, above-mentioned character zone is carried out the character recognition step once again of character recognition once again according to the combinations of directions different with the combinations thereof direction with above-mentioned first condition.Thus, can reduce the low situation of discrimination that causes owing to erroneous judgement to combinations of directions.
Description of drawings
Fig. 1 is the block diagram that the character recognition device of implementing character recognition method of the present invention is shown.
Fig. 2 is the process flow diagram that the embodiment 1 of character recognition method of the present invention is shown.
Fig. 3 illustrates will indulging of embodiment 1 to organize the concept map that first rule of horizontal group of character changed in character.
Fig. 4 illustrates will indulging of embodiment 1 to organize the concept map that second rule of horizontal group of character changed in character.
Fig. 5 illustrates will indulging of embodiment 1 to organize the concept map that the three sigma rule of horizontal group of character changed in character.
Fig. 6 is the concept map that first rule of the vertical group character of embodiment 1 horizontal group of character changed into is shown.
Fig. 7 is the concept map that second rule of the vertical group character of embodiment 1 horizontal group of character changed into is shown.
Fig. 8 is the process flow diagram that the embodiment 2 of character recognition method of the present invention is shown.
Fig. 9 is the process flow diagram of processing that the step S813 of Fig. 8 is shown.
Figure 10 is the process flow diagram that the embodiment 3 of character recognition method of the present invention is shown.
Figure 11 is the process flow diagram that the embodiment 4 of character recognition method of the present invention is shown.
Figure 12 is the process flow diagram that the embodiment 5 of character recognition method of the present invention is shown.
Embodiment
[embodiment 1]
With reference to the accompanying drawings the embodiment 1 of character recognition method of the present invention is illustrated below.
Fig. 1 is the block diagram that the character recognition device of implementing character recognition method of the present invention is shown.Fig. 2 is the process flow diagram that the embodiment 1 of character recognition method of the present invention is shown.Fig. 3 illustrates will indulging of embodiment 1 to organize the concept map that first rule of horizontal group of character changed in character.Fig. 4 illustrates will indulging of embodiment 1 to organize the concept map that second rule of horizontal group of character changed in character.Fig. 5 illustrates will indulging of embodiment 1 to organize the concept map that the three sigma rule of horizontal group of character changed in character.Fig. 6 is the concept map that first rule of the vertical group character of embodiment 1 horizontal group of character changed into is shown.Fig. 7 is the concept map that second rule of the vertical group character of embodiment 1 horizontal group of character changed into is shown.
In Fig. 1, character recognition device has the scanner 105 as the parts of the image input that will comprise character zone.Be stored in RAM103 from the image of scanner 105 inputs through bus 200.What be connected with bus 200 has CPU101, ROM102 and a memory storage 104.CPU101 controls whole character recognition device according to being stored in the program behavior among the ROM102.RAM103 is used as and is used for perform region of handling input image data etc.Memory storage 104 is memory units such as hard disk drive, is used for depositing file and picture, the alternative document of process object.In addition, character recognition device also can directly carry out the character recognition processing to the data that comprise the image that is stored in the document areas in the memory storage 104.
In addition, what be connected with bus 200 also has printer 106 and network I/F107, and printer 106 is with the character string of recognition result and is embedded into the printing unit of the outputs such as file and picture of watermark.Network I/F107 is to be used for and will to send to the communication component of external mechanical as the compressed image of discerning result.
The 108th, the keyboard that is used for carrying out input operation.The 109th, as the display of output block.
In Fig. 2, utilize following each step to carry out the character recognition method of in the character recognition device of Fig. 1, implementing.Such as, realize this character recognition method by carrying out the program that is stored in ROM102 or the memory storage 104 by CPU101.
Step S201:CPU101 is to numbering " i " initialization that is used for carrying out the character zone of character recognition below step S206 and forward step S202 to.
Step S202:CPU101 imports the image that comprises character zone from memory storage 104 or scanner 105, is read into RAM103.Forward step S203 afterwards to.
Step S203 (character extraction step): CPU101 carries out zone identification to the image of importing in step S202, extract character zone.Forward step S204 afterwards to.
Step S204 (combinations of directions determination step): CPU101 to the character area that extracts, utilizes the combinations of directions judging part in step S203, write across the page or the vertical judgement of writing.The combinations of directions judging part is made of CPU101, the ROM102 of Fig. 1, RAM103 etc.If can judge combinations of directions, adopt which kind of method can.Such as, also can adopt the decision method described in the patent documentation 1.
Step S205:CPU101 judges whether the processing in alphabet zone is all finished.If there is untreated character zone, just enter step S206, and the processing in the alphabet zone is when having finished all, just according to such end process.
Step S206:CPU101 selects i character zone as the character recognition object.
Step S207:CPU101 to i character zone, carries out character recognition according to the combinations of directions of judging in step S204, obtain the character code and the data such as line number, number of characters of recognition result.When generating character code, CPU101 calculates the similar degree of each character and the mode standard of identification dictionary.
Step S208 (condition criterion step): CPU101, the character identification result of inspection step S207.That is, CPU101, the data of inspection recognition result judge whether the judged result of combinations of directions is appropriate.
The inspection of character identification result, rule is according to the rules carried out.In the present embodiment, use following rule.In Fig. 3～Fig. 7, row is represented with rectangle frame, and character is represented with circle.
[for the 1st rule of vertical group character zone]
As shown in Figure 3, being judged to be the vertical character zone of organizing is 1 row (CL1 represents with symbol).And, be 1 character when (C1 represents with symbol), CPU101, judgement should be changed into horizontal group (representing to go R1, character TC1).This is that the major part of the character zone of 1 row, 1 character is horizontal group a empirical law.
[for the 2nd rule of vertical group character zone]
As shown in Figure 4, being judged to be the character zone that comprises multirow (totally 4 going for CL1, CL2, CL3, CL4 in the drawings) is vertical group.In this character zone, smaller or equal to 2, and line number is during more than or equal to 2 times of max number of characters at number of characters of each row, CPU101, and judgement should be changed into horizontal group with the judged result of the combinations of directions of this character zone.
In Fig. 4, under the state of vertical group, row CL1 comprises character C11 and C12 totally 2 characters.Row CL2 comprises character C21 and C22 totally 2 characters.Row CL3 comprises character C31 and C32 totally 2 characters.Row CL4 comprises character C41 totally 1 character.Line number 4 is 2 times of max number of characters 2.
If it is changed into horizontal group, then character C11, C21, C31, C41 become character TC11, TC12, TC13, the TC14 of capable R1.And character C12, C22, C32 become character TC21, TC22, the TC23 of capable R2.
[for the 3rd rule of vertical group character zone]
As shown in Figure 5, the character zone with multirow (totally 6 going for CL1, CL2, CL3, CL4, CL5, CL6 in the drawings) is judged to be vertical group.In this character zone, each the row number of characters at 3 characters interior, the long L of row of the row of 3 characters (being CL1 in the drawings) is smaller or equal to 1.5 times of line width W, and, line number is during more than or equal to 2 times of max number of characters 3, CPU101, judgement should be changed into horizontal group with the judged result of the combinations of directions of this character zone.
In Fig. 5, under the state of vertical group, row CL1 comprises character C11, CL12 and C13 totally 3 characters.Row CL2 comprises character C21 and C22 totally 2 characters.Row CL3 comprises character C31 and C32 totally 2 characters.Row CL4 comprises character C41 and C22 totally 2 characters.Row CL5 comprises character C51 and C52 totally 2 characters.Row CL6 comprises character C61 and C62 totally 2 characters.Line number 6 is 2 times of max number of characters 3.And the long L of row of the capable CL1 of 3 characters is in 1.5 times of its line width W.
If it is changed into horizontal group, then character C11, C21, C31, C41, C51, C61 become character TC11, TC12, TC13, TC14, TC15, the TC16 of R1.And character C12, C22, C32, C42, C52 and C62 become character TC21, TC22, TC23, TC24, TC25 and the TC26 of R2.And character C13 becomes the character TC31 of capable R3.
The illustrative processing of Fig. 5, such as, when character " three " being identified as the character zone of horizontal group 3 row etc. effectively.
[for the 1st rule of horizontal group of character zone]
As shown in Figure 6, being judged to be the character zone that comprises multirow (totally 4 going for R1, R2, R3, R4 in the drawings) is horizontal group.In this character zone, smaller or equal to 2, and line number is during more than or equal to 2 times of max number of characters at number of characters of each row, CPU101, and judgement should be changed into vertical group with the result of determination of the combinations of directions of this character zone.
In Fig. 6, under horizontal group state, row R1 comprises character C11 and C12 totally 2 characters.Row R2 comprises character C21 and C22 totally 2 characters.Row R3 comprises character C31 and C32 totally 2 characters.Row R4 comprises character C41 totally 1 character.Line number 4 is 2 times of max number of characters 2.
If it is changed into vertical group, then character C11, C21, C31 and C41 become character TC11, TC12, TC13 and the TC14 of capable CL1.And character C12, C22, C32 become character TC21, TC22, the TC23 of capable CL2.
[for the 2nd rule of horizontal group of character zone]
As shown in Figure 7, the character zone that will comprise multirow (in the drawings for R1, R2, R3, R4, R5, R6 totally 6 row) is judged to be horizontal group.In this character zone, each the row number of characters in 3 characters, the long L of row of the row of 3 characters (being R1 in the drawings) is smaller or equal to 1.5 times of line width W, and, line number is during more than or equal to 2 times of max number of characters 3, CPU101, judgement should be changed into the recognition result of the combinations of directions of this character zone vertical group.
In Fig. 7, under horizontal group state, row R1 comprises character C11, C12 and C13 totally 3 characters.Row R2 comprises character C21 and C22 totally 2 characters.Row R3 comprises character C31 and C32 totally 2 characters.Row R4 comprises character C41 and C42 totally 2 characters.Row R5 comprises character C51 and C52 totally 2 characters.Row R6 comprises character C61 and C62 totally 2 characters.Line number 6 is 2 times of max number of characters 3.And the long L of row of the capable CL1 of 3 characters is in 1.5 times of its line width W.
If it is changed into vertical group, then character C11, C21, C31, C41, C51 and C61 become character TC11, TC12, TC13, TC14, TC15 and the TC16 of CL1.And character C12, C22, C32, C42, C52 and C62 become character TC21, TC22, TC23, TC24, TC25 and the TC26 of CL2.Character C13 becomes the character TC31 of capable CL3.
Step S209:CPU101 just enters step S210 for should change combinations of directions the time judging in step S208, and just jumps to step S211 for should not change combinations of directions the time judging in step S208.
Step S210 (character recognition step once again): CPU101 discerns processing with the combinations of directions different with original combinations of directions, and the result is replaced.By discerning processing once again, can separate character and half-angle character correction recognition result to " three ", " い " etc.That is, can under the condition desirable, carry out character recognition at combinations of directions.
Step S211: " 1 " is added to " i " goes up and turn back to step S205.
According to embodiment 1, can reduce the low situation of discrimination that causes owing to erroneous judgement to combinations of directions.
[embodiment 2]
With reference to the accompanying drawings the embodiment 2 of character recognition method of the present invention is illustrated below.
Fig. 8 is the process flow diagram that the embodiment 2 of character recognition method of the present invention is shown.Fig. 9 is the process flow diagram of processing that is shown specifically the step S813 of Fig. 8.
According to embodiment 1, the character zone that changes combinations of directions is discerned processing once again.Relative therewith, in embodiment 2, omit identification once again and handle.That is,, when the similar degree of recognition result is higher than institute's definite value (below be called the 2nd condition), only change combinations of directions with regard to end process for the row of number of characters " 1 ".Thus, for the low character zone of the necessity of discerning processing once again, can omit identification once again and handle.
In Fig. 8 because step S801～S807 is corresponding with step S201～S207 of Fig. 2, so its explanation is omitted to being illustrated below the step S808.
Step S808 (the 2nd condition criterion step): CPU101 for i character zone through character recognition, judges whether the max number of characters in 1 row is 1.Max number of characters in 1 row is 1 and similar degree when being higher than institute's definite value, just should omit identification processing once again, forwards step S813 to.Max number of characters in 1 row is not 1 or similar degree when not being higher than institute's definite value, just forwards step S809 to.
Step S809:CPU101, the same condition of step S208 of utilization and Fig. 2 is checked recognition result.
Step S810:CPU101, the same with the step 209 of Fig. 2, judge in step S809, whether to be judged as to change combinations of directions.Being judged as to change combinations of directions the time, just enter step S811, and, just jump to step S812 being judged as should not change combinations of directions the time.
Step S811:CPU101, the same with the step S210 of Fig. 2, carry out character recognition once again with different combinations of directions and handle.
Step S812:CPU101 is added to " i " with " 1 " and goes up and turn back to step S805.
Step S813:CPU101 is the row of 1 character for max number of characters, omits the processing that identification is once again handled, and forwards step S812 to.
In Fig. 9, the processing of step S813 has following each step.
Step S901:CPU101 judges whether handle the character that obtains by the character recognition of step S807 might become " one " and wait owing to the different character code of combinations of directions.Such as, if vertical group is the possibility height of " " of Chinese figure.But, if horizontal group, might be " one " or the negative sign "-" of Chinese figure, the instability of character recognition increases.Like this, because combinations of directions difference and character code not simultaneously, just forward step S902 to, be when not being subjected to the character of influence of combinations of directions, then to jump to step S904.
Step S902:CPU101, the character of the occasion by the combinations of directions that will detect in step S804 compare with mode standard at the character of the occasion of the combinations of directions different with it and identification dictionary and calculate similar degree.So, both similar degrees are estimated.That is, since combinations of directions difference and character code not simultaneously, with reference to this character code.Such as, when being rotated 90 °, "-" (macron) just become taeniae, and the possibility that becomes unsuitable recognition result is very high.Therefore, if "-" (macron) is primary about similar degree, so just must adopt the processing of the character of other character codes that are predetermined (such as, " " of Chinese figure or the like) or deputy similar degree.
Step S903:CPU101 selects suitable character according to the evaluation result of step S902.
Step S904:CPU101 changes into different combinations of directions with character zone.Enter step S812 afterwards.
[embodiment 3]
With reference to the accompanying drawings the embodiment 3 of character recognition method of the present invention is illustrated below.
Figure 10 is the process flow diagram that the embodiment 3 of character recognition method of the present invention is shown.
In embodiment 3, be formation except embodiment 1, after identification is handled once again,, judge what person better (below's be called the 3rd condition) processing at recognition result that initial identification is handled with between through the recognition result that identification is handled once again.
In Figure 10, step S1001～S1010 is corresponding with step S201～S210 of Fig. 2, so its explanation is omitted to being illustrated below the step S1011.
Step S1011 (the 3rd condition criterion step): CPU101 to the character identification result on the initial combinations of directions (step S1007) with between the character identification result on the different combinations of directions (step S1010), judges which recognition result is better.If the result who judges is that the recognition result of step S1010 is better, just judges and adopt this result.On the other hand, if the recognition result of initial combinations of directions is better, just judges and adopt initial combinations of directions.Suppose that the decision condition that uses is an identical conditions with what use in step S1011, just judge by identification once again and handle in step S208, improve in the occasion that is not suitable for this decision condition.Handling by identification once again, with the occasion of the term harmonization of step S208, if average similar degree differs when equaling greater than predetermined value, just be judged as improvement.
Step S1012:CPU101 in the inspection of step S1011, judges whether to judge and should adopt initial combinations of directions.In the time should adopting initial combinations of directions, just forward step S1013 to.When judgement should not adopted initial combinations of directions, just forward step S1014 to.
Step S1013:CPU101 adopts the initial combinations of directions of judging at step S1004 (corresponding with the step S204 of Fig. 2).
Step S1014:CPU101 is added to " i " with " 1 " and goes up and turn back to step S1005.
In embodiment 3, confirm combinations of directions again by twice, can judge according to tighter combinations of directions and carry out character recognition.
[embodiment 4]
With reference to the accompanying drawings the embodiment 4 of character recognition method of the present invention is illustrated below.
Figure 11 is the process flow diagram that the embodiment 4 of character recognition method of the present invention is shown.
In embodiment 4, be formation except embodiment 1, also to the result of determination of initial combinations of directions, judge the whether high judgement of possibility of erroneous judgement.So, only when the possibility of by mistake discerning is high, check recognition result.
In Figure 11, step S1101～S1107 is corresponding with step S201～S207 of Fig. 2, so its explanation is omitted to being illustrated below the step S1108.
Step S1108:CPU101, by at image to process object, when carrying out the character recognition processing, judgement (the step S204 of step S1104, Fig. 2) result of initial combinations of directions is carried out statistical treatment one by one or the judged result of combinations of directions that the character recognition in past is handled is carried out statistical treatment, obtain the probability of erroneous judgement.Whether CPU101 utilizes the probability of judging by accident, judges the result of determination of the combinations of directions in step S1104, be the high result of determination of erroneous judgement possibility.When the possibility of judging by accident is high, just forward step S1109 to.On the other hand, when the possibility of judging by accident is not high, just forward step S1112 to.
Compare when being higher than setting such as, the erroneous judgement probability when in step S1104, the being judged to be vertical group erroneous judgement probability when being judged to be horizontal group, just only to a vertical result of determination of organizing, the processing of execution in step S1109～S1111.Judgement in step S1108 depends on the combinations of directions method of discrimination that adopts in step S1104.That is, because when the discriminating direction method of step S1104 has deviation on the erroneous judgement probability, only to the high direction determining processing of execution in step S1109～S1111 as a result of erroneous judgement probability, so can handle expeditiously.
Step S1109:CPU101, the character identification result of inspection step S1107.Promptly check the data of recognition result, judge whether the result of determination of combinations of directions is appropriate.In the present embodiment, condition of using in step S1109 and step S208's is identical conditions.
Step S1110:CPU101 judges when combinations of directions should change in step S1109, just enters step S1111.On the other hand, when judging that combinations of directions should not change, just jump to step S1112.
Step S1111:CPU101, the identification of carrying out on the combinations of directions different with original combinations of directions is handled, to handle the original recognition result of recognition result displacement that obtains by this identification, so, handle by carrying out identification once again, can separate character and half-angle character correction recognition result " three ", " い " etc.That is, can under the condition desirable, discern processing at combinations of directions.
Step S1112:CPU101 is added to " i " with " 1 " and goes up and turn back to step S1105.
In embodiment 4 because only to the possibility of mistake identification the recognition result when high check, unwantedly confirm again to handle so can omit, can make the processing high speed.
[embodiment 5]
With reference to the accompanying drawings the embodiment 5 of character recognition method of the present invention is illustrated below.
Figure 12 is the process flow diagram that the embodiment 5 of character recognition method of the present invention is shown.
In embodiment 5, allow result of determination not clear on initial combinations of directions, only in not clear occasion, check recognition result.
In Figure 12, step S1201～S1203 is corresponding with step S201～S203 of Fig. 2, so its explanation is omitted to being illustrated below the step S1204.
Step S1204:CPU101 to the character zone that extracts, utilizes the combinations of directions judging part in step S1203, carry out horizontal group or the vertical judgement of organizing.In this judges, allow vertical any not clear judgement of writing, write across the page.
Step S1205:CPU101, the same with the step S205 of Fig. 2, judge whether that the processing in alphabet zone all finishes.If there is untreated character zone, just enter into step S1206, and if the alphabet zone all processing finish, just according to such end process.
Step S1206:CPU101 selects i character zone as the character recognition object.
Step S1207:CPU101 judges whether the judgement in step S1204 is not clear.When being not clear, just enter into step S1208, the same with step S206～S210 of Fig. 2, discern the processing of identification once again of handling and changing combinations of directions.On the other hand, when being judged as vertical group or horizontal group, just enter into step S1213.
Step S1208:CPU101 to i character zone, carries out frequency than higher combinations of directions, such as, suppose it is " horizontal group ", character recognition, obtain the character code and the data such as line number, number of characters of recognition result.
Step S1209:CPU101 checks the character identification result in step S1208.Promptly check the data of recognition result, judge whether horizontal group supposition is appropriate.In the present embodiment, the decision condition that uses in step S1209 is identical with condition in step S208.
Step S1210: be judged as in step S1209 in the time of should changing combinations of directions, CPU101 just enters into step S1211.On the other hand, being judged as combinations of directions should not be changed the time, just enter into step S1212.
Step S1211:CPU101 changes into combinations of directions " vertical group ", discerns processing once again.
Step S1212:CPU101 is added to " i " with " 1 " and goes up and turn back to step S1205.
Step S1213:CPU101 carries out character recognition according to the combinations of directions of judging and handles in step S1204, enter into step S1212.
In embodiment 5, allow the judgement that combinations of directions is not clear, and, check to comprise the tight character recognition that identification is once again handled only to the not clear character zone of combinations of directions.Thus, can improve discrimination, and, there is unmatchful combinations of directions to judge the high character zone of precision, omit identification once again and handle, can improve processing speed.
In addition, in the above embodiment, be to suppose that carrying out character recognition method of the present invention with the character recognition device of Fig. 1 describes.But, it is self-evident can using multi-purpose computer to carry out character recognition method of the present invention.
Using multi-purpose computer to carry out in the character recognition method of the present invention, the computer executable program that needs to comprise the program code of each step that is used for making multi-purpose computer execution character method of identification is read into multi-purpose computer.
Be used for making the program of multi-purpose computer execution character method of identification, can or from the readable medium of multi-purpose computer, read in from the built-in ROM of multi-purpose computer, or read in from server etc. by network.
In addition, the spirit and scope of the present invention are not limited to specific explanation of the present invention and accompanying drawing.All corrections of the content of narrating in the scope to claim of the present invention and change all belong to interest field of the present invention, and this should it will be appreciated by those skilled in the art that.
According to the present invention, can reduce the low situation of discrimination that causes owing to erroneous judgement to combinations of directions.
Claims (12)
1. character recognition method, its formation comprise that the image to input carries out zone identification and step that character zone is extracted; Judge the combinations of directions determination step of the combinations of directions of the above-mentioned character zone that extracts; And the combinations thereof direction that obtains according to the utilizing above-mentioned judgement character recognition step that above-mentioned character zone carried out character recognition;
It is characterized in that:
In this character recognition method, also comprise step:
Judge the whether consistent condition criterion step of first condition of the result utilize the character recognition that above-mentioned character recognition step obtains and regulation; And
When consistent, above-mentioned character zone is carried out the character recognition step once again of character recognition once again according to the combinations of directions different with the combinations thereof direction with above-mentioned first condition.
2. character recognition method as claimed in claim 1 is characterized in that the afore mentioned rules condition is the line number of character zone, the number of characters of each row, and the similar degree of the mode standard of character that identifies and identification dictionary.
3. character recognition method as claimed in claim 1 or 2, it is characterized in that its constitute also comprise judge above-mentioned character zone whether with the step of the 2nd term harmonization of regulation,
For with the character zone of above-mentioned the 2nd term harmonization, do not carry out above-mentioned condition criterion step and above-mentioned character recognition step once again.
4. character recognition method as claimed in claim 1 or 2 is characterized in that its formation also comprises,
After carrying out above-mentioned identification step once again, judge and utilize this 3rd whether consistent condition criterion step of the 3rd condition of the character identification result that obtains of identification step and regulation once again, and
With above-mentioned the 3rd term harmonization the time, adopt the step of the combinations of directions that in combinations thereof direction determining step, obtains.
5. character recognition method as claimed in claim 1 or 2 is characterized in that also comprising combinations of directions in above-mentioned condition, discern the high combinations of directions of possibility for mistake, carries out above-mentioned condition criterion step, above-mentioned character recognition step once again.
6. character recognition method as claimed in claim 1 or 2, it is characterized in that in combinations thereof direction determining step, also comprising the determination step of judging whether combinations of directions is failed to understand, being judged to be the combinations thereof direction when not clear, carry out above-mentioned condition criterion step and above-mentioned character recognition step once again.
7. character recognition device, its formation comprise that the image to input carries out zone identification and extracting said elements that character zone is extracted; Judge the combinations of directions judging part of the combinations of directions of above-mentioned character zone; And the character recognition parts that carry out character recognition according to the combinations thereof direction;
It is characterized in that:
In this character recognition device, comprising:
Judge the whether consistent condition criterion parts of first condition utilize character identification result that above-mentioned character recognition parts obtain and regulation; And
When consistent, above-mentioned character zone is carried out the parts of character recognition once again of character recognition once again according to the combinations of directions different with the combinations thereof direction with above-mentioned first condition.
8. character recognition device as claimed in claim 7 is characterized in that above-mentioned first condition is the line number of character zone, the number of characters of each row, and the similar degree of the mode standard of character that identifies and identification dictionary.
9. as claim 7 or 8 described character recognition devices, it is characterized in that its constitute also comprise judge above-mentioned character zone whether with the step of the 2nd term harmonization of regulation,
For with the character zone of above-mentioned the 2nd term harmonization, do not make the action of above-mentioned condition criterion parts and the above-mentioned parts of character recognition once again.
10. as claim 7 or 8 described character recognition devices, it is characterized in that it constitutes and also comprise judgement at the 3rd whether consistent condition criterion parts of the 3rd condition of the character identification result that utilizes above-mentioned identification component once again to obtain and regulation, and
With above-mentioned the 3rd term harmonization the time, adopt the parts of the combinations of directions that obtains by combinations thereof direction determining parts.
11., it is characterized in that in above-mentioned condition, also comprising combinations of directions as claim 7 or 8 described character recognition devices, discern the high combinations of directions of possibility for mistake, make above-mentioned condition criterion parts, the action of the above-mentioned parts of character recognition once again.
12. as claim 7 or 8 described character recognition devices, it is characterized in that combinations thereof direction determining parts judge that also combinations of directions is whether not clear, when judging that the combinations thereof direction is not clear, make the action of above-mentioned condition criterion parts and the above-mentioned parts of character recognition once again.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2003109423A JP4346942B2 (en) | 2003-04-14 | 2003-04-14 | Character recognition method |
| JP109423/2003 | 2003-04-14 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1551038A true CN1551038A (en) | 2004-12-01 |
| CN100401314C CN100401314C (en) | 2008-07-09 |
Family
ID=33470590
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2004100348162A Expired - Fee Related CN100401314C (en) | 2003-04-14 | 2004-04-14 | Character recognizing method |
Country Status (2)
| Country | Link |
|---|---|
| JP (1) | JP4346942B2 (en) |
| CN (1) | CN100401314C (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101650779B (en) * | 2008-08-11 | 2013-01-23 | 欧姆龙株式会社 | Character recognition device and method |
| CN106650714A (en) * | 2016-10-08 | 2017-05-10 | 迪堡金融设备有限公司 | Paper note serial number identification method and apparatus |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH10289289A (en) * | 1997-04-14 | 1998-10-27 | Ricoh Co Ltd | Character recognition method and computer readable record medium recording character recognition processing program |
-
2003
- 2003-04-14 JP JP2003109423A patent/JP4346942B2/en not_active Expired - Fee Related
-
2004
- 2004-04-14 CN CNB2004100348162A patent/CN100401314C/en not_active Expired - Fee Related
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101650779B (en) * | 2008-08-11 | 2013-01-23 | 欧姆龙株式会社 | Character recognition device and method |
| CN106650714A (en) * | 2016-10-08 | 2017-05-10 | 迪堡金融设备有限公司 | Paper note serial number identification method and apparatus |
Also Published As
| Publication number | Publication date |
|---|---|
| JP2004318346A (en) | 2004-11-11 |
| JP4346942B2 (en) | 2009-10-21 |
| CN100401314C (en) | 2008-07-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1269069C (en) | Symbol identifying device and method | |
| CN1752991A (en) | Apparatus, method and program for recognizing characters | |
| CN1139884C (en) | Method and device for information treatment and storage medium for storaging and impelementing said method program | |
| CN1752992A (en) | Character recognition apparatus, character recognition method, and character recognition program | |
| JP5262493B2 (en) | Image processing apparatus, image processing method, program, and computer-readable storage medium | |
| CN1991865A (en) | Device, method, program and media for extracting text from document image having complex background | |
| CN1258894A (en) | Apparatus and method for identifying character | |
| CN1877598A (en) | Method for gathering and recording business card information in mobile phone by using image recognition | |
| CN1453747A (en) | Cluster | |
| US20020015525A1 (en) | Image processor for character recognition | |
| CN1477852A (en) | Image processing equipment, image processing method and storage medium of image processing program | |
| CN1607542A (en) | Image processing apparatus, image processing method, program and storage medium | |
| CN1141666C (en) | Online character recognition system for recognizing input characters using standard strokes | |
| CN101061512A (en) | Authentication device, authentication method, and program | |
| JP2011008549A (en) | Image processor, image reader, multifunctional machine, image processing method, program, and recording medium | |
| CN1878231A (en) | Apparatus and method for detecting secure document | |
| CN1484810A (en) | Method and device for recognising manually traced characters on an input zone | |
| CN1367460A (en) | Character string identification device, character string identification method and storage medium thereof | |
| CN1174338C (en) | Character identifying method, program and recording media | |
| JP5672059B2 (en) | Character recognition processing apparatus and method, and character recognition processing program | |
| CN1551038A (en) | Character recognizing method | |
| US20110002018A1 (en) | Image processing apparatus, image processing method, and storage medium for eliminating blurring of scanned image | |
| CN1896997A (en) | Character string searching device and its program product | |
| CN1308889C (en) | Method, device and storage medium for character recognition | |
| CN1092822C (en) | Full address reading apparatus |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20080709 Termination date: 20200414 |