A Method to Segment the Marked Regions in Document Images

It´s very popular to mark the concerned regions on a document using a mark pen. How to detect the marked regions in the scanned document images is a difficult problem. In this paper, an approach to solve the above problem is proposed. First, the color pixels and run-length are detected. And then, the mark-lines of document can be attained by using run-length. Finally, the mark-lines to get the marked regions are analyzed. Experimental results show the proposed method has high accuracy. Furthermore, it can be applied to many practical problems.