摘要 |
PURPOSE:To improve the reading performance of an optical character reader by performing majority processing based on only the result of identification of the same digit through a digit matching process so that a correct answer is obtained despite omission of characters. CONSTITUTION:The character strings forming a one digit are segmented out of binarized signals stored in a screen memory 8 via a one-digit segmenting circuit 9'. Then the (x) coordinates (a) are detected for each character. While the binarized signals of segmented one digit are stored in a one-digit memory 10. The result of identification obtained via a one-digit segmenting circuit 11, a one-character memory 12 and a character identifying result circuit 13 is stored in an identifying result buffer 14' together with the coordinates (a). Then the character images on a form 3 are taken in again by an image sensor 6 and the result of identification of the sensor 6 is stored in an identifying result buffer 15' as well as in a identifying result buffer 26' together with the coordinates (a). A digit matching processing part 18 matches the digits of the result of identification based on the coordinates (a). Then a majority processing part 17 performs decision by majority to decide one result of identification to one digit. This result is sent to a host computer B.
|