基于新型改进Canny算法的铁轨边缘检测研究

Research on rail edge detection based on new improved Canny algorithm

针对目前铁轨异物目标检测中,铁轨边缘提取困难,边缘检测结果不清晰,完整性较差等问题,提出一种基于新型改进Canny算法的铁轨边缘检测方法。首先,使用中值滤波代替高斯滤波,进行铁轨图像平滑处理;然后,将图像转换到频域空间进行频域滤波,增加高频信息,抑制低频信息,锐化边缘,减小照度不均匀造成的影响;其次,使用线性插值的方式改进非极大值抑制,进一步增强的边缘检测的准确性;再次,使用Ostu法得到阈值,增强铁轨边缘检测的自适应性;使用改进的双阈值边缘连接算法,减少边缘漏检;最后,通过计算边缘点连通数量和起始点距离,消除连通域和起始点距离过小的边缘,减小干扰。实验表明,改进后的算法进行铁轨边缘检测结果具有更好的连续性、完整性和清晰度。

In view of the difficulties of rail edge extraction,unclear edge detection results and poor integrity in rail foreign object detection,a rail edge detection method based on new improved Canny algorithm is proposed・Firstly,median filter is used instead of Gaussian filter to smooth the rail image;Then,the image is converted to the frequency domain space for frequency domain filtering,which increases the high frequency information,suppresses the low frequency information,sharpens the edge and reduces the influence of uneven illumination;Secondly,the linear interpolation method is used to improve the non maximum suppression to further enhance the accuracy of edge detection;Thirdly,Ostu method is used to get the threshold to enhance the adaptability of rail edge detection;The improved double threshold edge join algorithm is used to reduce the missing edge detection;Finally,by calculating the number of connected edge points and the distance from the starting point,the edge with too small distance between the connected domain and the starting point is eliminated,and the interference is reduced.The experimental results show that the improved algorithm has better continuity,integrity and clarity when it is used for rail edge detection.