基于双目视觉的汽车缝隙尺寸检测技术研究

Research on Automobile Gap Size Detection Technology Based on Binocular Vision

为了实现对汽车表面缝隙尺寸的快速检测以保证整车的使用性能,提出了一种基于最小外接矩形的缝隙尺寸检测方法。对双目相机进行标定并获取缝隙图像之后,对缝隙图像进行去噪、边缘检测的预处理,通过最小外接矩形定位缝隙并进行固定点匹配,使用最小二乘法实现对特征点的三维重建,最终实现了对汽车表面缝隙尺寸的检测。以大众汽车的车身缝隙为检测对象,检测数据分析基于VS2017和OpenCV实现,检测结果的相对误差为2.4%。试验结果表明:提出的缝隙检测方法有望满足对于汽车车身缝隙尺寸检测精确、高效的要求。

In order to realize rapid detection of gap size on the vehicle surface and ensure the performance of the vehicle,a detection method of the gap size based on the minimum external rectangle is proposed.After the binocular camera is calibrated and the slit image is obtained,the slit image is preprocessed by denoising and edge detection.The slit is located by the minimum external rectangle and fixed point matching is carried out.The three-dimensional reconstruction of feature points is realized by using the least square method.Finally,the detection of the automobile surface slit size is realized.The detection data analysis is based on VS2017 and OpenCV,and the relative error of detection result is 2.4%.The experimental results show that the proposed method is expected to meet the requirements of accurate and efficient detection of automobile body gap size.