An iterative approach to detect region boundary eliminating texts from scanned land map images

Land map images are a complex document as it contains different kinds of information such as place names, the border of states or country etc. As texts may have different intensity value and size, it is not easy to extract these texts from land map images. This paper designs an iterative approach to extract texts from land map images. The approach deals with finding outlines using a mask in an iterative way to separate text and outlines. The proposed approach is tested on a set of land map images and the results are encouraging.