JPH04220787A - Character reader - Google Patents

Abstract

PURPOSE:To prevent a word not registered in a word dictionary from being erroneously checked with another word by preventing the word not registered in the word dictionary from being checked in the word dictionary. CONSTITUTION:According to a control method designated by a control method destination part 17, a data control part 13 transmits recognition result received from a recognition part 10 to a word checking part 14 or correction part 12. By preventing the word not registered in the dictionary from being checked in the dictionary, time for checking a character not required for checking can be saved. Furthermore, since the erroneous checking can be prevented, the character erroneously checked cannot be missed and the time for correcting the erroneously checked character can be saved.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character reading device that recognizes input image data as characters and compares the recognition results with a word dictionary.

0002

2. Description of the Related Art As described in Japanese Patent Application Laid-Open No. 1-82192, when a recognition result is compared with a word in a word dictionary, a conventional device is used to compare characters that can be matched with words and characters that cannot be matched with words. The display method is changed to make it easier to distinguish between the two.

0003

[Problem to be Solved by the Invention] In the above-mentioned conventional technology, the operator pays attention to and corrects the characters that could not be matched with words, and the characters that could be matched with words are assumed to have been correctly matched. The problem is that characters that have been word-matched may be overlooked. In particular, in the case of words that are not registered in the word dictionary, there is a high possibility that the words will be erroneously matched.

An object of the present invention is to prevent an operator from overlooking erroneously verified results.

[0005]

[Means for Solving the Problems] In order to achieve the above object, it is specified in advance whether or not to check with a dictionary for each predetermined unit.

[0006]

[Operation] By specifying in advance whether or not to check with a dictionary for each predetermined unit, and by preventing words that are not listed in the dictionary from being checked against the dictionary, words that are not listed in the dictionary can be mistakenly compared to other words. to prevent it from being matched against.

[0007]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be explained in detail below with reference to the drawings. FIG. 1 is a block diagram of a character reading device according to an embodiment of the present invention.

The recognition unit 10 reads the characters written on the form as image data, recognizes each character as a character code, and sends the recognition results to the recognition result output unit 11. The recognition result output unit 11 receives 1 sent from the recognition unit 10.
The recognition results for each character are determined in a predetermined unit specified by the operator (for example, line by line, field by field, form by form, etc.)
The data is sent to the data control unit 13 every time. The data control section 13 sends the recognition result for each predetermined unit to the correction section 12 or the word matching section 14 according to a control method specified in advance by the control method specification section 17 . The word matching section 14 is connected to the data control section 1
The recognition results sent from 3 are compared with the word dictionary 15,
The collated results are sent as recognition results to the data control section 13, and the data control section 13 sends them to the correction section 12. The correction unit 12 displays the sent recognition results, and the operator corrects the recognition results if necessary. Then, the recognition result is sent to the output section 16 according to the operator's instructions. Further, when the operator instructs to perform word matching on the recognition results, the correction section 12 sends the recognition results to the data control section 13, and the data control section 13 sends the recognition results to the word matching section 14. The output unit 16 outputs the recognition result sent from the correction unit 12 to a storage device such as a magnetic disk. In the control method specifying section 17, the operator can specify the control method in the data control section 13 in advance.

FIG. 2 shows the flow of recognition results controlled by the data control section. The flow includes a flow 21 from the recognition result output section 11 to the correction section 12, and a flow 21 from the recognition result output section 11 to the correction section 12.
→ Word matching section 14 → Flow 22 of correction section 12 and correction section 1
There are three types of flows: 2 -> word matching section 14 -> correction section 12 flow 23.

The control method designation section 17 can designate a combination of flows of recognition results. There are four combinations as follows:
The case of flow 21 is defined as control method 1. Flow 21 and Flow 2
Case 3 is defined as control method 2. The case of flow 22 is assumed to be control method 3. The case of flow 22 and flow 23 is assumed to be control method 4. Control method 1 is specified when the word written in a predetermined unit is not registered in the word dictionary. Control method 2 is specified when a part of the word written in a predetermined unit is registered in the word dictionary. Control methods 3 and 4
is specified when most of the words written in a predetermined unit are registered in the word dictionary.

Next, an example of actually inputting a form into a character reading device and having the characters read will be explained.

FIG. 3 shows an example of a form to be input into a character reading device. In this example, the form consists of a name (Kanji) field 31, a name (furigana) field 32, a postal code field 33, an address field 34, a product code field 35, and a product name field 36. Become.

FIG. 4 is an example of a screen displayed when specifying a control method. FIG. 5 is an internal table. First, the operator selects a predetermined unit in the control method specifying section 17 before reading the form, and in this example, since the field unit is selected, the operator specifies the control method for each field, and selects a word dictionary. When performing a check, specify the dictionary to check using the check method. In this example, the control method for fields 31 and 32 is control method 3,
The matching method is to match the name, that is, a dictionary in which the name is registered. The control method for fields 33 and 34 is control method 4, and the verification method is address. The control method for the field 35 is control method 1, which does not perform a checking method, that is, does not check against a dictionary. The control method in field 36 is control method 2, and the verification method is product name. After specifying the control method and verification method, the control method specifying section 17 converts it into an internal table and sends it to the data control section. Items that use the same control method and verification method are grouped. In this example,
Field 31 and field 32 belong to group A. Field 33 and field 34 belong to group B, field 35 belongs to group C, and field 36 belongs to group D.

FIG. 6 is a flowchart showing the operation during reading. FIG. 7 is a flowchart at the time of correction. The data control unit 13 performs the operation shown in the flowchart of FIG. 6 when the recognition result is sent from the recognition result output unit 11 (when reading characters), and when the read data is sent from the correction unit 12 ( (at the time of correction), the operation shown in the flowchart of FIG. 7 is performed.

When the form shown in FIG. 3 is input to the character reading device, the recognition unit 10 reads the characters on the form as image data, cuts out each character one by one, recognizes it as a character code, and sends it to the recognition result output unit 11. The recognition result output unit 11 is
In this case, since the predetermined unit is a field unit, the recognition result for each field is sent to the data control unit 13 (step 6
1). After acquiring the recognition results for all fields on the form (step 62), the data control unit 13 refers to the table in FIG. 5 and performs processing for each group. For groups whose control method is 3 or 4 (step 64), data of all fields of the group is sent to the word matching unit 14 (
Step 65), and verification is performed according to the verification method (Step 66). Next, based on the matching result, the recognition result of the group is replaced with the matching result (step 67). After the processing of one group is completed, the processing of the next group is executed (step 68). When processing for all groups is completed (step 63), all data is sent to the correction unit 12 (step 69).
.

The correction unit 12 displays the recognition result received from the data control unit 13, and when the correct answer is the first candidate, confirms it, and when the correct answer is the second or lower candidate, selects the candidate.
The recognition result is sent to the output unit 16. When a correct answer is not obtained and it is not checked against the word dictionary, the recognition result is corrected to obtain the correct answer and the recognition result is sent to the output unit 16. When checking against a word dictionary, the dictionary is checked without modification or with partial modification. For example, if the operator selects field 36 in FIG.
When performing word matching on the data, the operator issues an instruction to the correction unit 12 in a predetermined manner (for example, by positioning the cursor on the screen in the field 36 and pressing a predetermined key). The correction unit 12 corrects the field 36
The data control unit 13 is instructed to apply word matching to the data. The data control unit 13 searches the table of FIG. 5 for the field 36 taught by the modification unit 12, and obtains the group to which the group belongs (step 71). When the control method of the acquired group is 2 or 4 (step 7
2) Send the data of all fields of the group to the word matching unit 14 (step 73) and match according to the matching method (step 74). After verification, modify the verification results in section 1
2 (step 75).

Even if control method 2 or 4 is specified, if recognition is performed correctly, word matching can be stopped by performing a predetermined operation. Conversely, even if it is not supposed to match words, it is possible to perform word matching later.

In this embodiment, the control method is specified in advance from the control method specifying section, but the control method is specified by reading the content of data written on the form (for example, identification ID number, etc.). It is also possible for the unit 17 to make a judgment and switch the control method for each predetermined unit.

[0019]

[Effects of the Invention] Words not listed in the dictionary are not checked against the dictionary, thereby saving time for checking characters that do not need to be checked. Furthermore, since there is no chance of erroneous matching, characters that are erroneously matched will not be overlooked, and there is an effect that it will not take time to correct the erroneously matched characters.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a character reading device according to an embodiment of the present invention.

[Figure 2] Diagram showing the flow of recognition results controlled by the data control unit

[Figure 3] Diagram showing an example of a form

[Figure 4] Diagram showing an example of screen display when specifying a control method

[Figure 5] Diagram showing an internal table

[Fig. 6] Flowchart showing processing when reading by a character reading device

[Figure 7] Flowchart showing processing when correcting the character reading device

[Explanation of symbols]

10... Recognition unit, 11... Recognition result output unit, 12... Correction unit,
13... Data control unit, 14 Word matching unit, 15... Word dictionary, 16... Output unit, 17... Control method designation unit.

Claims (1)

[Claims] 1. A recognition unit that recognizes input image data and outputs a character code as a recognition result; a word matching unit that matches the input recognition result with a word dictionary and outputs the matching result as a recognition result; a control method designation part that designates whether or not to compare the recognition results with a word dictionary for each predetermined unit; , and a data control unit that sends the recognition result received from the recognition unit to the word matching unit when the recognition result received from the recognition unit is checked against the word dictionary, and to the correction unit when the recognition result is not checked against the word dictionary, according to the instruction specified by the control method specifying unit. A character reading device characterized by: