JPH04302070A - Character recognizing device - Google Patents

Abstract

PURPOSE:To enhance the efficiency of correcting work by displaying characters selected from plural character patterns having degrees of resemblance higher than a prescribed value and characters, which are judged to be absent in a vocaburary dictionary, in distinction from each other. CONSTITUTION:When input characters are converted from picture data to character data, words which may be erroneous are displayed as candidate words requiring correction. In this case, characters selected from plural patterns resembling them at the time of using a recognition dictionary 5 to perform pattern collation and characters whose collation results do not exist in a vocaburary dictionary 7, that is, which are judged to be unknown words are displayed on a display part 2 in distinction from each other. Thus, an operator refers to this displayed state to efficiently perform post-editing like correction.

Description

[Detailed description of the invention]

[Purpose of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character recognition device.

0002

2. Description of the Related Art With recent advances in electronic document processing, character reading devices have been developed that read documents printed on paper and convert them into electronic media. A conventional character recognition device uses a reading unit to read a manuscript in which various documents are written. At this time, the reading section does not capture the document as a collection of characters, but reads it as image data consisting of a collection of dots. Then, the character recognition unit cuts out an arbitrary range from the read image data and performs pattern matching processing with standard character patterns stored in a recognition dictionary. As a result, if the character recognition unit can match the characters, this character is output to the output unit and recorded in the recording unit. By performing these processes, characters on the document are converted into codes within the apparatus. However, in these processes, ambiguity occurs between similar characters such as the number '0' and the alphabet 'O', the lowercase '1' and the uppercase 'I', and the number '1'. In some cases, there may be some residual error, and it was necessary to teach the operator about this part.

Furthermore, even if the character recognition unit determines that there is no ambiguity, there may be reading errors due to blurred characters in the document. Therefore, as a post-processing step, the read characters are grouped into words and it is determined whether the word exists in the vocabulary dictionary (hereinafter referred to as unknown word determination), and if it does not exist, it is a reading error. It was necessary to instruct the operator that this was a possibility. Both cases of reading errors and cases where ambiguity remains are called recognition errors.

However, in conventional character recognition devices, even if the operator is instructed to identify a candidate for a recognition error location, the operator must check whether the candidate location has been determined to be ambiguous by character recognition or whether it is an unknown word. It was not possible to obtain information as to whether the word was determined to be an unknown word or not, and the work of correcting the recognition results was burdensome. In addition, the system sometimes points out words that are correctly recognized because a vocabulary dictionary is not maintained, or points out words that are correctly recognized because they are ambiguous.

[0005]

[Problems to be Solved by the Invention] In this way, in conventional character recognition devices, even if the operator is instructed on candidates for recognition error locations, it is difficult to determine whether the error is due to ambiguity in character recognition or because it is an unknown word. Since the operator does not know what is happening, there is a problem in that the subsequent correction work is a heavy burden. Furthermore, there is a problem in that even though the word is correctly recognized, it is not possible to prevent the word from being recognized incorrectly due to an underdeveloped vocabulary dictionary or poor judgment of ambiguity.

The present invention has been made in consideration of such circumstances, and its purpose is to provide a character system that can indicate to an operator the cause of a recognition error in a candidate for a recognition error location and improve the efficiency of correction work. The purpose of this invention is to provide a recognition device. [Structure of the invention]

[0007]

[Means for Solving the Problems] A character recognition device according to the present invention includes an input section for inputting characters as an image pattern;
A character recognition unit that compares the input image pattern with character patterns in a pre-stored recognition dictionary and selects characters whose similarity is greater than or equal to a predetermined value; It is equipped with an unknown word determination unit that determines whether or not the unknown word exists in a character string of a stored vocabulary dictionary, and a display unit that displays the obtained character, and the character recognition unit detects characters whose similarity is equal to or higher than a predetermined value. In the case of a character string containing a character selected from among multiple patterns, an identifier A is assigned and stored, and in the case of a character string determined by the unknown word determination unit to be an unknown word that does not exist in the vocabulary dictionary, an identifier B is assigned and stored. However, when a predetermined condition is met, a character string and an identifier corresponding to the character string are displayed.

[0008]

[Operation] According to the present invention, words that may have made an error when converting input characters from image data to character data are displayed as candidate words that require correction. When performing pattern matching using the lexicon, whether there were multiple similar patterns (A) or whether it was determined that the matched result does not exist in the vocabulary dictionary (B) is displayed or stored. This allows the operator to perform post-editing such as correction while referring to the displayed state.

[0009]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows a block diagram showing the configuration of a character recognition device according to an embodiment of the present invention.

In FIG. 1, an input section 1 is capable of inputting commands necessary for the operation of the character recognition device of the present invention and inputting correction characters for the recognition results.
A keyboard, mouse, touch panel, etc. are used. The display unit 2 displays commands and characters input from the input unit 1, and displays the reading results from the reading unit 3, the intermediate status and recognition results of character recognition by the character recognition unit 4, which will be described later.
It is capable of displaying the contents of the instruction by the correction-required candidate word/phrase instruction unit 8, and a bitmap display or the like is usually used. It is also possible to divide the screen into two, displaying the image read by the reading section on one side, and displaying the corresponding characters recognized by the device on the other side. The reading unit 3 is for reading a document, and uses, for example, a scanner equipped with a line image sensor. In addition,
The resolution of this scanner determines the types of characters that can be read. The character recognition unit 4 recognizes the image data obtained as a result of the reading unit 3 reading the original as characters. When the reading section 3 reads a document, it does not read each character, but sequentially scans the document starting from one end, and obtains image data by representing each point in binary form, indicating whether or not there is information. For this reason, the higher the resolution, the finer the document will be read, and the more complex characters can be read. The image data provided to the character recognition section 4 can be obtained not only by being read by the reading section 3 but also by inputting an image file 12 or the like. Then, the character recognition unit 4 performs a pattern matching process on an arbitrary range of the image data with a standard character pattern recorded in advance in the recognition dictionary 5. This pattern matching process is performed on all of the image data, and if a standard character pattern with a degree of similarity equal to or higher than a predetermined value matches, it is recognized as a character corresponding to the image data. The unknown word determination unit 6
It is determined whether or not the results recognized by the character recognition unit 4 described above are recorded in the vocabulary dictionary 7 on a word-by-word basis. The candidate word/phrase instruction unit 8 that requires correction is used to identify ambiguous characters or unknown words, such as when the read pattern matches multiple patterns through pattern matching by the character recognition unit 4, or when there is a reading error in which the read pattern does not match any pattern. The word determined to be an unknown word by unit 6 is determined to be likely to have been erroneously recognized in character recognition, and is used to instruct the operator about the character or word. The identifier display section 9 is for distinguishing and displaying whether the word or phrase pointed out by the candidate word/phrase requiring correction section 8 has been determined by the character recognition section or the unknown word determination section. The recording unit 10 is
The image data read by the reading unit 3 and the character recognition unit 4
It records the intermediate and final results of character recognition. The control section 11 controls each section and mediates data.

FIG. 2 is a flowchart showing the overall processing flow of the character recognition device according to the present invention. In order to digitize a manuscript as character data, first, the manuscript is read by the reading section 3 (S201). Next, the character recognition unit 4 performs character recognition on the read result (S202). At this time, if there are words or phrases that remain ambiguous in the recognition results, they are designated as A. For example, a flag SP-F indicating ambiguity is set and stored in the recording unit 10 together with the phrase. Next, the unknown word determining unit 6 performs unknown word determination on the data for which character recognition has been completed (S203). When multiple patterns with similarity equal to or higher than a predetermined value appear during character recognition, it may be possible to determine unknown words only for the pattern with the highest degree of similarity, or it may be possible to determine unknown words for all patterns that appear. good. At this time, if there are words that are unknown words, they are designated as B. For example, a flag undef-F indicating that the word is an unknown word is set and stored in the recording unit 10 together with the word. Now that the recognition has been completed and the locations that may require correction have been determined, this data is displayed on the display unit 2 for confirmation and correction by the operator (S204). Then, in accordance with the operator's instructions, candidate words requiring correction are pointed out and an identifier for the word is displayed (S205). At this time, the first candidate with the highest degree of similarity and its identifier are displayed, but if there are second or subsequent candidates, these may also be displayed together with the identifier using the next candidate key or the like. The operator corrects the document data from the input unit 1 while referring to the instructions and display (S206), and obtains the final result. In addition,
If there is no pattern with similarity greater than a predetermined value during character recognition, unknown word determination and character data cannot be displayed, so only the ambiguous identifier may be displayed, or a "reading error" may be displayed separately. May be displayed.

To explain the process of FIG. 2 in more detail, there are several variations, examples of which are shown in FIGS. 3 to 5. FIG. 3 is a flowchart for explaining the operation of the step of pointing out a candidate word/phrase requiring correction (by the correcting candidate word/phrase instruction unit 8) and displaying an identifier for the word/phrase (by the identifier display unit 9). If you want to point out candidate words that need correction and display the identifier for those words, first,
Extracts the first word of the document to be processed (S301, S3
02) Let it be D. Determine whether or not D includes a character that is ambiguous as a result of character recognition (whether or not D exists in word group A in the recording unit 10) (S303);
If it is included, D is a candidate word requiring correction, and an instruction to that effect is given to the display unit 2 (S304). At the same time, an identifier indicating that this instruction contains ambiguous characters as a result of character recognition is displayed (S305
). If it is determined in step 303 that no ambiguous characters are included, the next step is to determine whether D is an unknown word (whether D exists in word group B in the recording unit 10). A determination is made as to whether or not (S306). If it is an unknown word, D is a candidate word requiring correction, and an instruction to that effect is given to the display unit 2 (S307). At the same time, an identifier indicating that this instruction is because D is an unknown word is displayed (S308). When D is displayed on the display unit 2 together with the identifier as a candidate word requiring correction, the operator looks at this and, if correction is necessary, makes the appropriate correction (S206 in FIG. 2). In this way, after pointing out the candidate word that needs to be corrected for D and displaying the identifier for that word, the operator is asked whether to point out the next candidate word that needs to be corrected (S309). repeats these steps until the next candidate is found or the data is exhausted. In the process shown in FIG. 3, for words with ambiguity, the final result is obtained only by confirmation and correction by the operator, regardless of whether they are unknown words or not.

The process shown in the flowchart of FIG. 4 differs from the process of FIG. 3 in that it is determined whether or not an ambiguous word is an unknown word. Similar to FIG. 3, the words and phrases of the document to be processed are taken out in order and designated as D. If it is determined that D is ambiguous (included in word group A) as a result of character recognition (S403 yes), D is displayed as a candidate word requiring correction along with an identifier indicating that it is due to ambiguity (
S404, S405). The operator checks and corrects the D thus displayed (S406), and then judges whether or not this D is an unknown word (S407). If D is determined to be unambiguous (S403 n
o) is also judged as an unknown word (S407), and if it is judged as an unknown word (included in word group B) (S408).
yes), D is displayed as a candidate word requiring correction along with an identifier indicating that it is an unknown word (S4
08, S409). The operator checks and corrects D thus displayed (S410).

In the process shown in FIG. 4, if a candidate word that is determined to require correction due to ambiguity is actually a recognition error, the operator corrects it and then checks whether it is an unknown word. Therefore, for words that have been corrected by the operator, the correction result D' may be applied to the unknown word determination section 6 at step S407 to determine whether or not the correct word D' is an unknown word. By doing this, words that have been determined as unknown words do not contain ambiguity, so only words that are purely unknown words are processed in steps S408 and S409.
will be displayed at Therefore, if the operator issues an instruction to register the displayed unknown word in the vocabulary dictionary 7 at the time of S410, the vocabulary dictionary will be enriched, and from now on, if the correct word is not found due to a deficiency in the vocabulary dictionary, the operator can issue an instruction to register the displayed unknown word in the vocabulary dictionary 7. It will no longer be considered as a candidate for correction.

The process shown in the flowchart of FIG. 5 allows the operator to set conditions for indicating that a word is a candidate word requiring correction. Similar to FIG. 3, the words and phrases of the document to be processed are taken out in order and designated as D. If D is included in both word groups A and B (it is ambiguous and is also an unknown word) (S503), D is included in word group A but not in word group B (even if there is ambiguity, it is an unknown word). ) (S504), D is included in word group B but not included in A (there is no ambiguity but it is an unknown word) (S505)
, D is not included in the word group A or B (there is no ambiguity and it is not an unknown word) (S506), and the processing is performed separately. S503 is highly likely to be a recognition error. S504
, S505 may be judged as candidates requiring correction due to incorrect ambiguity judgment or insufficient number of words registered in the vocabulary dictionary, even though they are correctly recognized. Therefore, the operator sets whether or not to display these as correction-required candidates, and only when the setting to display them is made (S510,
S512) Display (S511, S513). The operator looks at this display and makes appropriate corrections (Fig. 2S
206), register the unknown word in the vocabulary dictionary if necessary. In the case of S503, which is an ambiguous and unknown word, when displaying this as a candidate word that requires correction, whether to display its identifier as ambiguous (S507)
, the operator can set the priority order of whether to display the word as an unknown word (S508). As described above, the settings of the logical conditions to be displayed as candidate words requiring correction and the settings of the priority order of identifiers may be changed depending on the appearance rate of ambiguous words.

FIG. 7(a) shows an example of the state of the display section 2 when performing the above-described processing. Candidate words that require correction in the recognition result data are displayed in reverse video, and if the identifier corresponding to this word D is ambiguous, for example, "Ambiguity +
", if there is no word, "ambiguity -" (or nothing is displayed), if it is an unknown word, "unknown word +", otherwise "unknown word -" (or nothing is displayed), and so on.

In the process of FIG. 5, both the ambiguity and unknown word flags are displayed at the same time, but in the process of FIG. Do not show. Further, in the process of FIG. 4, after the user confirms and corrects the word displayed as "ambiguity+", an unknown word flag is displayed.

In the above embodiment, the ambiguity of character recognition is first checked, and then unknown words are checked, but this order may be reversed. In addition, in the above example,
Although character recognition and unknown word determination were performed separately in advance, they may be performed simultaneously or may be performed in conjunction with displaying the results. For example, in FIG. 4, S403
Character recognition is performed on the word D extracted at step , its ambiguity is determined, and if there is ambiguity, it is displayed (S404, S40
5) correction (S406), and then in S407 it is checked whether this D is an unknown word that is not in the vocabulary dictionary, and if it is an unknown word, it is displayed and corrected, and after this series of processing, the next word is changed to D.
It can be transformed in various ways, such as by taking it out as

FIG. 6 shows an example of a configuration in which the above-described character recognition device is used as a translation device. Reading section 603
Then, a document written in the first language (for example, English) is read as image information, and the character recognition unit 604 compares the pattern of this image information with the standard character pattern of the recognition dictionary 605 to recognize it as a character code. . The unknown word determination unit 606 performs unknown word determination based on whether a phrase formed by arranging these character codes exists in the vocabulary dictionary 607 . The data that has been recognized and determined to be an unknown word is sent to a translation unit 612 and converted into a second language (for example, Japanese) text using a translation dictionary 613 . The vocabulary dictionary 607 is created by compressing the first language data in the translation dictionary 613 so that it can be searched quickly. Therefore, when adding unknown words, the unknown words and their corresponding translations are registered not only in the vocabulary dictionary but also in the translation dictionary.

The storage section 610 stores image information from the reading section 603, recognition results and progress from the character recognition section 604, judgment results from the unknown word judgment section 606, and the translation section 612.
Stores the translation results etc. Candidate words/phrases requiring correction section 60
Based on an instruction from an input unit (for example, a keyboard) 601 and based on the data stored in a storage unit 610 , a display unit 602 indicates candidate words that need to be corrected as explained in connection with the character recognition device described above. and identifier display section 609
Performed against. The input section 601 is used by the operator to correct the indicated candidate word/phrase requiring correction and to give instructions to other sections.

[0021] In this translation device, the character recognition unit 604
, before the data processed by the unknown word determination unit 606 is translated, the user makes corrections via the input unit 601 while looking at the display on the display unit 602, thereby eliminating recognition errors in the character recognition device. Data translation unit 612
(a) Manual mode, in which the entire process from reading to translation is performed without human intervention, and the data that has been processed by the translation unit 612 is displayed on the display unit 602, allowing the user to make corrections (b) Automatic mode There is.

FIG. 7 shows an example of indication of candidate words requiring correction on the display section 602. FIG. 7(a) shows the case in manual mode (the display is the same as the display on the display unit 2 of the character recognition device in FIG. 1), and FIG. 7(b) shows the case in automatic mode. In either case, the candidate words that need to be corrected are shown in sequence, e.g. highlighted, and the identifier corresponding to the highlighted word is placed somewhere on the same screen to identify the ambiguity and/or unknown word. The flag is shown as +. This display method may be performed in other ways, such as changing the color of the inverted display and using it as an identifier. (b) In the automatic mode, both the character-recognized data and its translation results are displayed on separate screens. In Fig. 7(a), regarding the word claims in the manuscript whose characters have been recognized, the ambiguity + indicates that there were other similar characters during pattern matching.
The unknown word - indicates that ms is a word in the dictionary. In FIG. 7(b), va in the manuscript (original text) whose characters have been recognized is
This shows that the word riuus (in the original manuscript it was various) has ambiguity and is an unknown word. In the translated text, the portion corresponding to the unknown word variuus is indicated by, for example, a highlighted blank space.

(a) In the manual mode, data without recognition errors is translated, which has the effect of simplifying post-editing to correct the translation results. Furthermore, since words appearing together with an unknown word identifier can be registered in the dictionary before translation, the efficiency of translation is improved. Moreover, the work of correcting the recognition result at the stage of recognition is also made more efficient as explained in connection with the above-mentioned character recognition device. (b) In the case of automatic mode, conventionally, the reason why translation was not possible was (1) a reading error in the translation source data. (2) It was an unknown word that was not registered in the dictionary, (3) It was neither (1) nor (2), but the context could not be understood. When translation failed, the operator was unable to distinguish the cause. This translation device solves this problem. Taking Figure 7(b) as an example, if the ambiguity identifier + is the cause (1), if the unknown word identifier + is the cause (
2), and if it is neither of these, but the translation has not been completed, it is the cause (3), which has the effect that the operator can easily guess the reason why the translation has not been completed.

By the way, conventional character recognition devices are provided as independent systems dedicated to character recognition. For this reason, when translating a character code that has been recognized as a character and stored, for example, on a disk of the device, the code is stored on a floppy diskette or the like and then registered in the translation system. For this reason, operators are required to acquire knowledge such as the file names of documents whose characters have been recognized, how to save them on floppy diskettes, code systems, etc. This operation is complicated and requires a great deal of effort on the part of the operator. .

[0025] Also, checking whether the character code string that has been recognized is incorrect is carried out using a spell check dictionary (the vocabulary dictionary 7,607 in the above embodiment) that is provided in advance in the character recognition device. However, there is a problem in that it is not possible to determine whether a recognized character code string exists in a dictionary used by a system that performs the next desired process, for example, a machine translation system.

[0026] Therefore, a control means is provided on the character recognition device side so that the character code string can be passed to the next processing system simply by instructing the next processing method to be performed on the recognized character code string. A character recognition system that can reduce the operator's labor will be described as a second embodiment.

FIG. 8 is a block diagram showing a schematic configuration of a character recognition system according to the second embodiment. In FIG. 8, 81 is an input section, and input data of this input section 81 is sent to a character recognition control section 83. The input unit 81 is capable of inputting characters and various commands for controlling the character recognition system, and uses a keyboard, a mouse, etc. For example, an instruction to start inputting a document in the image input section 82 is displayed on the display section 89.
This is done by selecting the "Image reading" button with the mouse.

The character recognition control section 83 is for controlling the entire system, and here includes an input section 81, an image input section 82, an image storage section 84, a character recognition section 85, a character code storage section 86, and a character code storage section 83. It controls an editing section 87, a next processing control section 88, and a display section 89.

The image input unit 82 inputs one or more printed or handwritten manuscripts from a set state.
Optically reads the image on the paper. The read image is sent to the character recognition control section 83, subjected to compression, etc., and stored in the image storage section 84. If multiple pages are set, read in order up to the last page. The image input unit 82 is also used when rereading or inserting pages at the operator's request during recognition processing. The image storage unit 84 stores the read images page by page. The character recognition unit 85 recognizes the characters of the read image. The character recognition technique is as described in the first embodiment. The character code storage unit 86 stores the character code recognized by the character recognition unit 85. At this time,
It is stored in association with the image of the page read.

The character code editing section 87 includes the character recognition section 85
The character code recognized in is displayed on the display section 89, and editing such as insertion or deletion is performed according to the user's instructions. For example, it is possible to search for ambiguous characters among the characters recognized by the character recognition unit 85. At this time, a corresponding portion of the read image stored in the image storage section 84 is extracted and displayed in the vicinity of the ambiguous character. The character code string that has been edited by the character code editing section 87 is stored in the character code storage section 86 again. The display section 89 uses a device that can visually display information such as character reading images, such as a bitmap display.

[0031] Before passing the character code string that has undergone character recognition and editing to another processing system, the next processing control unit 88 receives instructions from the user regarding the processing to be performed on the character code string in the destination processing system. Select by. For example, when passing a character code string to a translation system, select the processing method, such as whether to simply register it as a document in the translation system, whether to register and translate it, or whether to print the translation result after translation. do. Furthermore, in the case of translation, information such as the name of the dictionary used for translation and the translation environment can be set.

Next, the operation of this system configured as described above will be explained. First, FIG. 9 is a state transition diagram for simply explaining the flow of processing in this character recognition system. 10 to 19 are screen examples for explaining the operation. FIG. 10 shows the initial screen of the character recognition system. The operator first inputs the document identification surface of the document to be read. Thereafter, when the "image reading" button is selected with the mouse, the image input section 82 starts operating (step 2).
01).

While the image input unit 82 is in operation, the read images shown in FIG. 11 are sequentially displayed on the screen. When a plurality of pages are present in the image input section 82, reading is repeatedly executed as long as there are pages (step 202). When image reading is completed, the process returns to FIG. 10 (step 203
).

[0034] When the initial screen is returned, the document identification name can be changed and an image of another document can be input. Also, if you are requested to input the next document without changing the document identification name, ask the operator if you want to add to the already scanned document or cancel the already scanned document and input a new document. do.

After inputting the image, if character recognition of the image is to be executed, the "recognition" button is selected with the mouse (step 204). The character recognition system of this embodiment recognizes the layout of the document before starting character recognition. Indicate to the operator whether there are any errors in the recognized layout. FIG. 12 shows an example of a screen that informs the operator of the results of layout recognition. On this screen, use the "Replace" button to change the order of blocks that were incorrectly recognized. You can also specify them one by one using the "Specify All" button. In addition, you can also specify the range for character recognition. Also, “delete” parts that do not need to be recognized.
You can also delete it with the button. For example, page numbers, headers, footers, etc. that do not require machine translation can be deleted or removed from the recognition range.

Furthermore, if the read image is not clear, select the "Reread" button (step 2).
05). With this operation, it is possible to re-input an image with changed reading conditions (shading, contrast, document position, etc.) from the image input section 82. Furthermore, pages that are omitted during the image input stage can be easily inserted by selecting the "insert page" button.

When the character recognition start button is selected in FIG. 12, character recognition starts in accordance with the recognized layout order. During recognition, the position being recognized is displayed on the right side of FIG. 13 (step 306). Additionally, the recognized character codes are displayed in sequence on the right side of the screen. When character recognition is completed, FIG. 14 is displayed. At this time, if the density of the original to be read is incorrectly specified or the original is placed at an angle, many characters will be erroneously recognized. In this case, by selecting the "Reread" button, the user can reread the document on the corresponding page (step 207).

Selecting the "Next Candidate" button searches for ambiguous characters in character recognition from the cursor position (
Step 208). When an ambiguous character is found, an image is displayed near the ambiguous character as shown in FIG. The operator corrects the characters while referring to the image. Here, if the character recognition unit 85 also performs unknown word determination, it is necessary to add the function of displaying ambiguous characters and unknown words with their respective identifiers, as described in the first embodiment. Can be done.

Furthermore, character code strings that do not exist in the dictionary used by the translation system can also be presented. That is, the vocabulary used in the above-mentioned unknown word determination and the dictionary used by the translation system are made to match, and phrases with the above-mentioned unknown word identifiers are presented. Here, by inputting the translation word to be translated for the character code string, the character code string and its translation word are registered in the dictionary used by the translation system. Furthermore, this character code string is also registered in a dictionary used for unknown word determination. However, while this feature is useful in translation systems, e.g.
It is not necessary when sending to another computer.

When editing for one page is completed, the "next page" button is selected (step 209). If there is a subsequent page image, layout recognition for the next page is started. If there is no subsequent page, return to FIG. 10. At this time, since character recognition for all pages has been completed, the "image input" and "recognition" buttons are displayed in a shaded state. This indicates that the next processing request (translation request in this embodiment) remains for the read manuscript indicated by the document identification name. Also, for example, when reading the image on page 5,
If character recognition and editing are executed up to page 3 and interrupted midway, only the "Image input" button will be displayed in a shaded state, and the "Recognition" and "Translation" buttons will not be shaded, indicating that recognition has not yet been completed. This will show you that the images you need remain. Furthermore, if the "Translate" button is selected while an image that requires recognition remains, the screen shown in FIG. 16 is displayed and a warning is issued to the operator.

When the "Translation" button is selected, FIG. 17 is displayed (step 311). On this screen, the operator selects the next processing content that the operator wants the processing system to pass the recognized data to. In the embodiment, (1) registering in a translation system, (2) translating after registration, or (
3) After translation, select which one to print by number.

[0042] When the processing content is selected and the "Confirm" button is selected, the character code string consisting of multiple pages indicated by the document identifier is combined into one and the processing content to be performed is notified to the translation system ( step 212). Then, return to the initial screen. The operations performed by the next processing control unit 88 while the screen of FIG. 17 is displayed will be described below.

FIG. 20 shows that after character recognition is completed, the next desired processing method is selected for the document in the character code storage section 86, and the character code strings stored on a page-by-page basis are grouped together for the next processing. Shows the flow of data when passed. In the embodiment, when a process number is selected on the screen shown in FIG. 17 and the "Confirm" button is selected, this process is started. FIG. 21 is a flowchart for explaining the operation when the above-mentioned "Confirm" button is selected. In step 701, it is determined whether a process number has been selected, and if not, the operator is notified. Next, the selected processing method is output as processing control data (step 702). In the embodiment, an identifier "ctl" is added after the document identification name.

Next, the recognized character code strings for each page stored in the character code storage section 86 are retrieved in order of page number and output as a single set of text data (step 703). In the embodiment, the document identifier is followed by the identifier "t".
xt".

The output processing control data and text data are moved to the next desired processing and predetermined area (steps 704 to 705). As an area for the next desired processing and data transfer, a common area of a disk or the like is used as a predetermined area, for example. Moreover, this common area is not an area within the system, but an area of another computer connected via a network or the like may be used.

Next, the other processing system that performs the desired processing waits for processing control data and text data to be output from the character recognition system to a predetermined area (step 706). When the processing control data is output to a predetermined area, the processing instructed by the processing control data is started. Alternatively, the process is started by determining whether the above data is output at regular intervals (steps 707 to 711).

For example, in machine translation, the processing method of processing control data and the type of dictionary used for translation are determined, and the text data input by the character recognition system is registered in the translation system, or the document registration and translation is performed. Or even automatically print the translated text. here,
Returning to the explanation of FIG. 9.

There is a function to display a list of documents in order to know the status of the documents that have been read (step 213). FIG. 18 shows an example screen of the document list. The document name of the read document is displayed together with the number of pages read and the number of pages for which recognition has been completed. If a document is selected in the display state of this document list, there is no need to input a document identification name (step 214). Additionally, by selecting the "delete" button, unnecessary documents can be easily deleted (step 215).

In addition, the dictionary to be used when sending the recognized character code string to the translation system for translation can be easily selected by selecting the "Dictionary Selection" and "Translation Environment Setting" buttons (step 216). FIG. 19 shows an example of a dictionary selection screen. In the reading, character recognition, editing, and next processing control described above, information is managed on a page-by-page basis, and the operation of the character recognition control section 83 that performs this will be described below. Figure 2
2 is an image storage unit 84 that stores images of multiple pages read by the image input unit 82, and a character code storage unit 86.
FIG.

FIG. 22 is an example of data correspondence using the document identification name input on the initial screen shown in FIG. A document having two document identification names "roubun" and "manual" is input from the image storage unit 84, and three pages of "ronbun" are being recognized as characters.

[0051] The data name in the example is given by adding a 3-digit sequence number in the order of reading after the document identification name, and then adding "rf" and "txt" as data identifiers.
etc. are added and shown. In FIG. 22, data with the identifier “rf” and “rounbun004 to 005
” and “patent001-004” indicate that the read image is written as is in a compressed state.Furthermore, “ronbun001-003” has already been recognized as a character and is stored in the character code storage unit as a character code. Therefore, it is erased from the image storage unit 84. In the embodiment, page images whose characters have already been recognized are erased, but they may be stored in the image storage unit 84 until recognition of all pages is completed. "tx" in the character code storage section 86
Data with the identifier “t”, “ronbun00
1 to 003" indicates that character recognition has been completed. In addition, "rounbun. ctl'' stores the next desired processing method, dictionary type, original text identification name, etc. selected in FIGS. 17 and 19.

FIG. 23 shows rereading, page insertion, and page insertion performed on the read image stored in the state shown in FIG.
It is a flowchart for explaining the operation of page deletion. In this case, the character recognition control unit 83 determines whether the processing request from the input unit 81 is a page-related process (step 101
~103). Processing requests other than page deletion, rereading, and page insertion are processed by other functions (step 104).

If it is determined in step 103 that the operator's request is to insert a page, first the image storage unit 8 is
A comparison counter for extracting the data identification name assigned to page 4 on a page-by-page basis is cleared (step 105). Next, the data identification name indicated by the comparison counter is retrieved from the image storage unit 84, and the document identification name (i) and page number (i
) (step 106).

Next, the document identification name (i) extracted in step 106 is compared with the document identification name being processed (step 107). If they do not match, the comparison counter is incremented by one because the document identifiers are different (step 110).
). If the document identification names match, the page number (i) is compared with the page number being processed (step 108).

If the page number (i) is the page being processed or is large, the page number of the data identification name is incremented by one (step 109). Thereafter, the comparison counter is incremented to determine whether there is another data identification name in the image storage section 84. If there is another data identification name, the process returns to step 106 (steps 110 to 111).

For example, in FIG. 22, if the document identifier being processed is "ronbun" and a page insertion request is made, character recognition has been completed for up to page 3, so page 4 is the corresponding page. do. Therefore, the data identifier is “ron
Bun. 004. rf” → “ronbun. 005. r
f”, “ronbun. 005. rf” → “ronbu
n. 006. rf”.

When the change of the corresponding data identification name is completed, the image of the page to be inserted into the area indicated by the data identification name being processed is read from the image input unit 82, and the next process (layout recognition in the embodiment) is started. (step 112)
. If it is determined in step 102 that the operator's request is for re-reading, there is no need to change the data identification name, so step 112 is executed.

If it is determined in step 101 that the operator's request is to delete a page, first, the image storage unit 8
4, the data identification name being processed and the corresponding read image are deleted (step 113).

Next, the data identification name is changed in the same way as when inserting a page. In the case of page deletion, the page number of the data identification name is decreased by one (step 118). Other steps 11
Steps 4 to 120 are the same as steps 105 to 111 for inserting pages, so their explanation will be omitted.

The gist of the second embodiment described above is to provide a means for reading an image of a printed or handwritten manuscript;
The present invention is equipped with a means for character recognition of a read image, a means for editing characters encoded by character recognition, and a means for instructing the contents of processing to be performed next on the edited character code string. It is a character recognition system that is characterized by a means for reading images of multiple pages of printed or handwritten manuscripts, a means for storing the read images of multiple pages page by page, and a means for reproducing page by page in the reading means. means for reading, inserting pages, or deleting pages; means for character recognition and editing of stored images of multiple pages on a page-by-page basis; The present invention is a character recognition system characterized by comprising means for combining character code strings into one.

[0061] Therefore, according to the character recognition system configured in this way, when images of a plurality of pages are sequentially read and character recognition and editing processing is performed, it is possible to perform character recognition and editing without being aware of the pages. image character recognition,
After the editing is executed, the recognized and edited character code strings are combined into one, and the file name, The next process can be executed without being aware of the code system, etc., and the labor of the operator can be significantly reduced.

[0062] In the above embodiment, the processing of the recognized character code string corresponds to an example of translation, but it may be transmitted to another computer. Furthermore, the layout information may be converted into a DTP document format. Furthermore, instead of inputting the image using a scanner, data transmitted by fax may be input. It is possible to perform character recognition on images input from a fax machine, translate the recognized character code string using a translation system, and even print the image.

[0063]

[Effects of the Invention] As explained above, according to the character recognition device according to the present invention, it is possible to indicate to an operator the reason for pointing out a recognition error candidate, and also to Since it is possible to set various conditions, great practical effects can be achieved, such as reducing the burden on the operator in editing after character recognition and improving work efficiency.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing a schematic configuration of a character recognition device according to an embodiment of the present invention.

FIG. 2 is a flowchart showing an overview of the operation of a character recognition device according to an embodiment of the present invention.

3 is a flowchart illustrating an example of the operation of the step of pointing out a candidate word/phrase requiring correction and displaying an identifier for the word/phrase in FIG. 2;

4 is a flowchart illustrating another example of the operation after the step of pointing out a candidate word/phrase requiring correction and displaying an identifier for the word/phrase in FIG. 2;

5 is a flowchart illustrating another example of the operation of the step of pointing out a candidate word/phrase requiring correction and displaying an identifier for the word/phrase in FIG. 2;

FIG. 6 is a block diagram showing a configuration example of a translation device using the character recognition device according to the present invention.

[Figure 7] (a) Display section 2 in Figure 1, (b) Figure 6
The figure which shows the example of the screen of the display part 602 in.

FIG. 8 is a block diagram showing the configuration of a character recognition system according to a second embodiment.

FIG. 9 is a state transition diagram for explaining the flow of processing in the second embodiment.

[Fig. 10] An initial screen diagram of this character recognition system.

FIG. 11 is a screen diagram of the present character recognition system during image input.

FIG. 12 is a screen diagram showing layout recognition results of the present character recognition system.

FIG. 13 is a screen diagram of the present character recognition system during character recognition.

FIG. 14 is a screen diagram showing character recognition results of the present character recognition system.

FIG. 15 is a screen diagram when searching for ambiguous characters.

FIG. 16 is a screen diagram when a warning is issued for a request for the next process (translation) when recognition of all pages has not been completed.

FIG. 17 is a screen diagram when the operator is asked to select the next processing content.

FIG. 18 is a screen diagram when displaying a list of scanned documents.

FIG. 19 is a screen diagram when the operator is asked to select a dictionary to be used during translation.

FIG. 20 is a diagram showing the flow of data when recognized data is passed to the next process.

FIG. 21 is a flow diagram showing the operation of the next processing control unit 88 that instructs the next processing system to perform processing on data that has been recognized.

FIG. 22 is a diagram showing the relationship between an image storage section 84 that stores images of multiple pages and a character code storage section 86.

FIG. 23 is a flow diagram showing the operation of the character recognition control unit 83 regarding reading, character recognition, storage, editing, etc. on a page-by-page basis.

[Explanation of symbols]

1,601...Input unit 2,602...Display unit 3,603...Reading unit 4,604...Character recognition unit 5,605...Recognition dictionary 6,606...Unknown word determination unit 7,607...Vocabulary dictionary 8,608... Candidate words/phrases requiring correction section 9,609...
Identifier display unit 10, 610... Storage unit 11, 611... Control unit 12... Image file 612... Translation unit 613... Translation dictionary 81... Input unit 82... Image input unit 83... Character recognition control unit 84... Image storage unit 85... Character recognition section 86...Character code storage section 87...Character code editing section 88...Next processing control section 89...Display section

Claims (1)

[Claims] Claim 1: An input means for inputting characters as an image pattern, and an image pattern inputted by the input means is compared with a character pattern in a first dictionary stored in advance, and the degree of similarity is equal to or higher than a predetermined value. a character recognition means for selecting characters of the character pattern, and determining whether a character string consisting of the characters obtained by the character recognition means exists in a character string of a second dictionary stored in advance. In a character recognition device comprising a determination means and a display means for displaying characters obtained by the character recognition means, when displaying characters by the display means, a character string consisting of the displayed characters is When the character recognition means includes a character selected from a plurality of character patterns with the degree of similarity equal to or higher than a predetermined value, and when the determination means determines that the character does not exist in the character string of the second dictionary. A character recognition device characterized by distinguishing and displaying.