基于边缘连接的线段检测方法

Line segment detection method based on edge linking

提出了一种基于图像边缘连接的线段检测方法。数字图像中的线段是由很多线段图元组成,在同一条线段中,它们的方向是相同的并且受到真实线段的方向约束。因此可以通过提取小的线段图元,然后分析它们的连接特性,进行线段的检测。首先用ADM算法对图像进行边缘检测,得到单像素宽的边缘;然后在4个方向上进行线段的检测;对初步得到的结果进行合并和验证,去掉短的线段,连接临近的线段。经过实验证明,该方法能够快速地提取图像中的线段,并且对噪声具有一定的鲁棒性。

A fast line segment detection method is proposed based on edge linking. In digital image, digital lines consist of many line segment elements. The directions of these elements are same and are inhibited by the slope of digital line. Thus the detection of line segment can be carried out through extracting line segment elements and analyzing their connection. First, the edge is detected by ADM algorithm, and then rough line segments are extracted in four directions. In the end, the obtained results are testified and the line segments with same slope and intercept are merged, some shorter line segments are deleted. Experiments show that the proposed method can fast extract line segments in complex image, and is robust to noise.