结合LTS和Hough变换的直线检测算法

Line Detection Algorithm Combining LTS with Hough Transform

为提高Hough变换检测直线的精度,提出一种结合Hough变换与截断最小二乘法的直线检测算法。利用Hough变换确定图像中直线所在的大致区域,提取候选区域内的特征点集,采用截断最小二乘法拟合得到精确的直线参数。实验结果表明,该算法的检测率和检测精度较高,对Hough变换的分辨率要求较低,整体空间开销较小。

A novel line detection method combining Hough Transform（HT） and Least Trimmed Squares（LTS） method is proposed for improving the accuracy of the Hough transform for line detection. Image regions where lines exist are obtained through Hough transform. The LTS method fitting is applied on sets of feature pixels extracted from the candidate regions to obtain accurate line parameters. Experimental results show that the proposed method gives higher detection rate and accuracy, and the requirements on the resolutions of Hough transform are reduced, thus reducing the overall memory cost of the algorithm.