基于主线段的签名下划线去除与笔画修复方法

Method of removing underlines and repairing strokes in signature image based on main line segment

手写签名图像中的下划线影响离线签名鉴伪准确率,为实现签名骨架图像中下划线的去除及笔画修复,提出基于主线段的直线段检测(MLSD)和限距修复(RDR)方法。首先将手写签名骨架化,通过Hough变换初步检测出签名骨架中的线段,然后基于主线段的信息从中检测下划线线段并进行像素标记,去除各下划线像素并在其8邻域内找到笔画起点和终点像素,计算各起点和终点像素距离,最终通过阈值判断进行笔画连接修复。实验结果表明,该方法可以有效检测出签名图像中的下划线,并对去除下划线后的笔画进行准确的修复。

Underlines in handwritten signature images would reduce the accuracy of offline signature authentication.In order to remove the underlines and repair the strokes in signature skeleton images,main line segment detection(MLSD)and restricted distance repair(RDR)are proposed in this paper.Firstly,the handwritten signature skeleton image was obtained and the line segments within the signature skeleton were detected preliminarily based on the Hough Transform algorithm.Then the underlined segments were determined according to the major segment information and were further labeled at the pixel level.Finally,the underlined pixels were removed,and the starting and destination pixels of all the stroke pixels were discovered in their 8 neighborhoods.The distance between the starting pixel and the destination pixel was calculated and then the stroke connection restoration could be completed using threshold judgment.Experimental results show that the proposed method can effectively detect the underlines and accurately repair the underlined strokes in the signature skeleton images.