摘要
针对已有的提取钢轨表面方法在提取钢轨表面时存在过提取和欠提取的缺点,提出了一种基于Radon变换的快速钢轨表面提取方法.首先,根据钢轨表面区域向非钢轨表面区域进行过渡时像素点的灰度值会产生突变,用Radon变换生成原图像的灰度投影曲线及其投影曲线的平均值;然后,以投影曲线与其平均值的交点来确定积分的初始位置和结束位置.由于钢轨表面的宽度是确定不变的,因此从初始位置开始且以钢轨表面的宽度为区间对投影曲线进行灰度积分,直到结束位置,在此区间所得到的最大积分区间为要提取的钢轨的表面区域.将仿真结果与其它方法相比较表明:该方法抗噪能力强,而且还能准确、快速的提取钢轨表面区域.
Aiming at the existing method for extracting rail surface, the defects of over extraction and under extraction are found when extracting the rail surface. A fast rail surface extraction method based on Radon transform is proposed. First of all,according to the obvious jump of the gray brightness of the rail surface area and the non rail surface area, the Radon transformation is used to generate the gray scale projection curve of the original image and the mean value of its projection curves. Then the initial position and end position of the integral are determined by the intersection of the projective curve and its mean value. Because the width of the rail surface is cer- tain,the initial position is taken as the starting point,and the width of the rail surface is taken as an interval to make gray integral for projective curves till the end position. During the period,the maximum gray value integral interval is the surface area of the rail. Compared the simulation re- sult with that of other methods, it is showed that the method is robust to noise and can extract the rail surface region accurately and quickly.
作者
冉建民
顾桂梅
周咏
RAN Jian-min GU Gui-mei ZHOU Yong(School of Automation and Electrical Engineering, Lanzhou Jiaotong University, Lanzhou 730070 ,Chin)
出处
《兰州交通大学学报》
CAS
2017年第4期152-157,共6页
Journal of Lanzhou Jiaotong University
基金
甘肃省科技计划资助(1508RJZA059)