基于结构光视觉传感的轮对踏面擦伤快速检测

Detection of Tread Surface Flats of Wheelsets Based on Structural Light Visual Sensing

研究一种基于结构光视觉传感的铁路车辆轮对踏面擦伤快速检测方法。通过控制轮对转动速度,并以结构光扫描轮对踏面及应用CCD(Charge Coupled Device)视觉传感器实时采集轮对踏面外形特征信息,经过图像数字滤波、基准校正、边缘识别和空间坐标变换等算法,建立踏面擦伤数字矩阵。在此基础上,进行踏面擦伤特征信号的分析和提取,实现轮对踏面擦伤最大深度及形状的非接触快速准确检测。本文给出了试验结果,并分析了踏面数字矩阵的建立过程和视觉传感器标定对擦伤测量的影响。试验证明本文方法实用、可靠,可有效提高轮对踏面擦伤测量精度和效率。

A new method based on the structured light vision sensing for fast detecting the tread surface flats of railway wheelsets is presented. The structured light is applied to scan the tread surface and form different shapes of light to indicate the tread surface contour while the turning speed of wheelsets is controlled by a wheelset measurement device. This changed light is collected by a CCD vision sensor and processed by the image processing techniques such as image digital filtering, calibration, edge recognition and coordinate transformation. Also, a digital matrix that reflects the tread surface profile is established based on the structured light. The characteristics of the flats of the tread surface can be analyzed and extracted. By using this new method, the flats of the tread surface can be non-contactly detected and measured in real time. The experimental results have been given. The influence of the digital matrix establishment of the tread surface and the calibration of the vision sensor on the measurement of the flats is analyzed. Experimental results have demonstrated that this new method for measuring the concave depth and shape of the wheelsets flats is practical and reliable. The efficiency and the measuring precision of detecting the flats of the tread surface are promoted apparently.