JPS649577A - Optical character reader - Google Patents

Abstract

PURPOSE:To attain the read of mixed documents whose reference side varies by providing a means selecting the reference side on a document and a means calculating the segmentation position of characters from the detected reference side position. CONSTITUTION:When a reference side detection section 11 uses reference side information 141 of a document stored in a memory 14 so as to detect either right or left edge of the document as the reference side, the position information of the detected reference side is outputted as a data 112 and a control signal 111 is outputted. A segmentation position calculation section 12 calculates the data 112 being the position of the reference side and field position information 142 stored in the memory 14 and outputs the segmented position of the field character on a scan buffer 20 as a data 122 and outputs a control signal 121 representing it. A character segmentation section 13 segments a character pattern from the scan buffer 20 one by one character according to the segmentation position information of the data 122 to start the character recognition section 2. Thus, even when an attachment is added or no attachment is provided, the document is read.