基于聚类分析的精密零件测量方法

Large precision device measurement method based on cluster analysis

摘要:

针对目前大型精密零件自动化测量和自动化装配的迫切需求，提出一种精密零件轮廓自动化测量方法。首先利用激光跟踪仪建立全局坐标系，基于世界统一法对多个相机坐标系进行全局标定；采用像素点聚类的方式、漫水填充法等图像处理方法对存在重影以及环境光的激光光条中心进行精确提取，利用统计滤波方法对三维数据点云进行杂点的去除，结合粗匹配和精细匹配方法对多组相机下的激光光条进行三维空间点云融合，移动产品，对精密零件进行全产品扫描，最后，借助于射线法对精密零件轮廓进行超差预警。实验表明，该方法对大型零部件产品的测量误差标准差为0.25mm，15分钟可以完成20米长大型零部件产品的测量与分析，足够满足现有大型精密复杂产品的装配需求，突破了现有技术瓶颈，可以对精密零件轮廓进行快速精确三维重构，并且可以计算精密零件产品的结构化参数。

Abstract:

Aiming at the urgent demand of automatic measurement and automatic assembly of large precision device, an automatic measurement method of large precision device contour is proposed. First, the global coordinate system is established by laser tracker, multiple camera coordinate systems is calibrated based on the global calibration and all the camera coordinate systems can be transformed to the global coordinate system. The center of the laser strip is precisely extracted by the Image processing methods such as Clustering method, flooding method and so on, by which ghosting and ambient light can be overcome; Statistical filtering method is used to remove the clutter of 3d data point cloud, by combining the sample-consistent initial registration algorithm and the iterative nearest point matching method, the 3d point cloud fusion is carried out for the laser bars under multiple cameras. The whole product is scanned by moving the precision device, and when the contours of the precision device are out of tolerance ray method will give warning. The experiment shows that the measurement error standard deviation of this method for large parts is 0.25mm, and the measurement time is 15 minutes for 20 meters long parts, which is enough to meet the assembly requirements of large precision and complex products. This technology breaks through the existing technical bottleneck, it can carry out rapid and accurate 3d reconstruction of the contour of precision parts, and can calculate the structural parameters of precision parts.