JPH04242874A - Optical character reader - Google Patents

Abstract

PURPOSE:To automatically correct all similar reject characters in the same document only by correcting one word rejected due to the habit handwriting of a writer. CONSTITUTION:The optical character reader is provided with a scanner part 1 scanning the document optically, a recognition part 2 identifying characters in the image obtain from the scanner part 1, and a correction part 5 displaying the reading result and performing the correction by an operator. It also provided with a reject pattern collation storage part 3 storing unreadable characters (which is also called 'reject patterns') in the recognition part 2, a means 4 storing the correspondence between plural identical reject patterns in other places and the reject pattern characteristic, and a reject pattern correction part 5 detecting the reject pattern position having the same reject pattern characteristic with the reject pattern by the operator, and a conversion part 6 converting the character located in the reject pattern position into the corrected character.

Description

[Detailed description of the invention]

0001

[Field of Industrial Application] The present invention relates to an optical character reading device, and in particular, when reading a form filled in by a large number of unspecified persons and when a reject pattern, which is an unreadable character, is generated, the present invention can be used to read a form that is filled in by a large number of unspecified persons, and when a reject pattern that is an unreadable character is generated, The present invention relates to an optical character reading device that automatically corrects the same reject pattern appearing in other places in the same document by correcting only one character when a rejected pattern appears.

0002

[Prior Art] Conventionally, this type of optical character reading device corrects all characters of the reject pattern each time a reject pattern appears, or corrects the characters of the reject pattern by increasing the precision of the dictionary using an automatic learning function. I was trying to reduce the number of characters.

0003

[Problems to be Solved by the Invention] With the above-mentioned conventional optical character reading device, depending on the person writing, the same characters may be concentrated in a reject pattern, and all of them are subject to correction, so it is troublesome to correct them one by one. The disadvantage is that it occurs. In addition, improving dictionary accuracy through automatic learning is effective in reducing the number of rejected patterns for specified fillers, but for unspecified fillers, even similar patterns may have different meanings depending on the filler. , it has the disadvantage of leading to misreading.

0004

[Means for Solving the Problems] The optical character reading device of the present invention includes a scanner unit that optically scans a form, a recognition unit that identifies characters in an image obtained from the scanner unit, and a reading result. In an optical character reading device having a correction section for displaying and correcting by an operator, characters that cannot be read by the recognition section (referred to as a reject pattern)
a reject pattern matching/storage section for storing a reject pattern; and a reject pattern conversion unit that detects a reject pattern position having a pattern characteristic and converts a character at this reject pattern position into a corrected character.

[0005]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, the present invention will be explained with reference to the drawings. FIG. 1 is a block diagram showing one embodiment of the present invention. The scanner section 1 has a function of reading an image of a form and capturing characters on a form basis. The recognition unit 2 recognizes characters from the reading information given from the control unit 7 and passes the recognized character reading results to the correction unit 5. At this time, recognition unit 2
When a certain character is determined to be rejected, the reject pattern information and the position of the generated reject pattern character are passed to the reject pattern matching storage section 3. The reject pattern matching storage unit 3 compares the reject pattern feature given from the recognition unit 2 with the stored reject pattern features, and if there is the same reject pattern feature, it will notify its registration number, and if not, it will create a new reject pattern feature. and passes the new number along with the reject pattern position to the reject pattern feature/reject pattern position correspondence storage section 4. Reject pattern verification and erasure of the storage unit 3 is performed for each form. The reject pattern feature/reject pattern position correspondence storage unit 4 stores the reject pattern feature number obtained from the reject pattern collation/storage unit 3 in association with the reject pattern position. The correction unit 5 displays the read result, and the operator looks at the display and passes the correct value and position of the reject pattern corrected to the reject pattern conversion unit 6. The reject pattern conversion unit 6 searches the reject pattern feature/reject pattern position correspondence storage unit 4 from the reject pattern position given from the correction unit 5, determines the reject pattern feature, searches for the same reject pattern feature, and converts the reject pattern position. character is converted into the correct value given by the correction unit 5 and passed to the correction unit 5. Note that the control unit 7 is the recognition unit 2.
It provides reading information to the system and controls each part.

[0006]

As explained above, the present invention detects the same reject pattern position as the corrected reject pattern with the reject pattern matching storage unit, the reject pattern position and reject pattern feature correspondence storage unit, and the reject pattern position and reject pattern feature correspondence storage unit. It also has a reject pattern converter that converts characters into correct correct values, and if a character is rejected due to the wrong handwriting of the person filling it out, all similar rejected characters in the same form can be corrected by correcting one character. This has the effect of being automatically corrected.

[0007] Furthermore, since the automatically corrected data is deleted for each form, the same effect can be obtained even when a plurality of forms are filled out by unspecified persons.

[Brief explanation of the drawing]

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

[Explanation of symbols]

1 Scanner section 2 Recognition section 3 Reject pattern matching/storage section 4 Reject pattern feature/reject pattern position correspondence storage section 5 Correction section 6 Reject pattern conversion section 7 Control section

Claims (2)

[Claims] Claim 1: A scanner unit that optically scans a form, a recognition unit that identifies characters in an image obtained by the scanner unit, and a correction unit that displays the reading results and allows an operator to make corrections. In the optical character reading device, the recognition unit includes a reject pattern collation/storage unit that stores unreadable characters (referred to as reject patterns), and a reject pattern matching/storage unit that stores unreadable characters (referred to as reject patterns) in the recognition unit; means for storing the correspondence; and a reject pattern converting unit in which an operator detects a reject pattern position having the same reject pattern characteristics as a reject pattern in the correcting unit and converts a character at this reject pattern position into a corrected character. An optical character reading device comprising: 2. The optical character reading device according to claim 1, wherein erasing of the reject pattern matching storage section is performed for each predetermined form.