铁路货车车轮踏面伤损检测中剥离与擦伤定位方法

Locating Methods of Peeling and Flat Spots in Detection of Wheel Tread Damages of Railway Wagons

针对铁路货车车轮踏面伤损检测问题,设计铁路货车车轮踏面伤损动态检测系统,提出剥离和擦伤的定位方法。首先对车轮踏面图像进行平滑去噪;然后对图像进行基于平稳小波和Canny融合算法的边缘检测,并根据标准车轮的宽度确定踏面区域;最后,根据剥离和擦伤的不同特征,对剥离踏面区域图像进行基于分块思想和种子填充算法的剥离粗定位后,再进行基于Canny算法和跟踪法的剥离精定位,并给出基于踏面边缘线搜索的擦伤定位方法。实验结果表明:在铁路货车运行速度为0.5 km/h^10 km/h时,踏面剥离的定位准确率达到96.7%,定位精度达到3 mm;擦伤定位准确率达到97.8%;可满足现场动态检测的要求。

To solve the problems of wheel tread damages of railway wagons, a dynamic detecting system was developed and a new method for locating wheel tread peeling and flat spots was presented. Firstly, the wheel tread gray-image was preprocessed for smoothing denoising. Secondly, image edge detection was implemented with combination of the stationary wavelet transform and Canny algorithm and the tread region of the wheel was found on the basis of the width of the standard wheel. Finally, considering the different features of peeling and flat spots, approximate locating of peeling spots was achieved by block segmentation and the seed filling algorithm, and then precise locating of peeling spots was achieved by the Canny and tracking methods, also the locating method of wheel tread flat spots was proposed on the basis of searching of tread edge lines. Experimental results show that at the 0.5 km/h-10 km/h running speed of railway wagons, the correctness rate of peeling locating comes to 96.7% with 3 mm precision and the correctness rate of flat spot locating comes to 97.8% thus satisfying the requirements of dynamic detection on site.