基于边缘跟踪的直线特征检测算法

针对现有直线检测算法的缺陷，提出一种新的基于边缘跟踪的直线特征检测算法.算法对图像边缘进行提取，并在此基础上对边缘进行跟踪，从图像边缘直接提取出局部直线.设计了一种鲁棒直线拟合策略，并对拟合后的直线以一定的准则进行合并，准确获取图像中直线特征及端点参数，最终实现完整直线检测.典型图像直线检测结果表明，所提出的算法能准确检测出图像中的全局直线，并具有较快的运算速度，性能优于现有基于Hough变换和相位编组的直线检测算法.

Line feature detection algorithm based on edge tracking

For the disadvantages of the existing line detection algorithms, a novel line feature detection algorithm based on edge tracking was proposed. The local line was extracted from the image edge using the edge tracking method. A robust line fitting strategy was designed and the fitting line was merged by a certain criterion to obtain exactly the line feature and end parameters in images and realize the complete line detection. The line detection results for the representative images show that the proposed algorithm can detect the global lines in the images accurately and quickly, and its performance is superior to the existing line detection algorithms based on Hough transform and phase group.