基于Halcon深度学习和线扫相机标定的闸片剩余厚度测量

Residual Brake Pad Thickness Measurement Based on Halcon Deep Learning and Line Scan Camera Calibration

针对测量轨道车辆闸片剩余厚度的需求,本文提出了一种基于Halcon深度学习和线扫相机标定的闸片剩余厚度测量方法。首先,采用目标检测算法实现关键零部件定位,进而剪裁出闸片区域图像。在闸片区域图像进行语义分割,得到闸片的边缘,进而求得图像像素坐标系下的闸片厚度。进行线扫相机的标定,获得每个像素对应的世界坐标距离,进而求得世界坐标系下的闸片实际厚度。该方法对于不同车型和光照等环境变化具有自适应性,且取得较好的测量精度和实时性。

In order to meet the requirement of measuring the remaining thickness of rail vehicles brake pads,this paper proposes a method for measuring the remaining thickness of brake pads based on Halcon deep learning and line scan camera calibration.First of all,target detection algorithm is used to achieve the positioning of key components,and then the image of the brake pad area is cropped.Semantic segmentation is performed on the brake pad area image to obtain the edge of the brake pad,and then the thickness of the brake pad in the image pixel coordinate system is obtained.Calibration of the line scan camera is carried out to get the corresponding world coordinate distance of each pixel,and then to get the actual thickness of the brake under the world coordinate system.This method is adaptive to the environment changes of different vehicle types and illumination,and achieves good measurement accuracy and real-time performance.