Method and apparatus for recognizing character information at a variable

Abstract

A method and apparatus for recognizing character information in an original document with high recognition accuracy regardless of the size of a character to be recognized. An input image is processed by an image processing circuit and stored in an image recording unit. A character region in the recorded image data is detected by a recognition section, and the image corresponding to the detected character region is read with a predetermined magnification. Character recognition is performed on the image of the character region read at the predetermined magnification, in the recognition section. If the recognition accuracy does not satisfy a predetermined criterion, the same image is read again with a modified magnification wherein the magnification is determined on the basis of preset sizes and the size of the character pattern of a character to be recognized. The image obtained in this way is subjected again to character recognition processing until the recognition accuracy satisfies the predetermined criterion. A character pattern corresponding to the recognized character is generated, and recorded together with the image recorded in the image recording unit in an image recording section. Thus, an image is read with a magnification optimized depending on a character size of an original document image and on a recognition accuracy.