基于合并短线段的直线检测方法

Detecting Straight Line Based on Merging Short Line Segments

图像中的直线检测是检测复杂物体的基础,传统方法在检测较复杂的边缘图像时往往不太稳定,容易出现漏检和误检的现象。针对传统检测方法中的问题,提出一种基于合并短线段的算法。该方法对偏向水平的直线与偏向竖直的直线分开进行处理,首先将边缘图划分成若干区域,通过构造若干个短线段匹配矩阵以检测每一块区域中的短线段,再将属于同一条直线上的短线段合并,最后探测断裂的线段。该方法鲁棒性较好,改善了传统检测方法中的漏检和误检问题。另外,除能检测直线段外,该方法还能够检测曲线段。实验结果表明,相比霍夫变换算法,该算法检测直线的精度提高了3.9个百分点,达到35%。

Line detection in image is the basis of complex object detection.Traditional methods are often not stable when detecting com⁃plex edge images,and are prone to miss detection and false detection.Aiming at these problems in traditional detection methods,an al⁃gorithm based on merging short line segments is proposed.Firstly,the edge image is divided into several regions,and the short line segments in each area are detected by constructing several short line segment matching matrices;then,the short line segments belong⁃ing to the same line are merged;finally,the broken line segments are detected.The method has good robustness and improves the prob⁃lems of missing detection and false detection in traditional detection methods;In addition,in addition to detecting straight line seg⁃ments,this method can also detect curved lines.Experimental results show that,the accuracy of our algorithm is 3.9 percentage points higher than that of Hough transform algorithm,reaching 35%.