Method and device for detection and refinement of text-containing regions of an image

摘要

A method of selectively refining image corners is disclosed wherein a region may be scanned. The region may be classified as either text-containing or non-text-containing. It may be then determined whether the region is text-containing or non-text-containing. If the region is text-containing, the missing corners within the text-containing region may be identified. Then, the missing corners may be added to an edge map.