基于双目视觉的齿轮三维点云精确重构方法研究

Research on accurate reconstruction method of gear 3D point cloud based on binocular vision

基于双目视觉测量系统开展面向齿轮的三维轮廓点云精确重构测量方法研究。针对齿轮的初始三维点云存在的噪点问题,利用直通滤波和体素栅格滤波去除离群点和冗余点;采用移动最小二乘法平滑加贪婪三角化重构算法进行了齿轮重构。搭建了双目视觉齿轮测量系统实验平台,通过测量实验对比分析了齿轮的三维重构效果及尺寸测量精度。结果表明:采用移动最小二乘法平滑加贪婪三角化重构后的齿轮三维模型比直接重构的齿轮表面光滑,边缘清晰,齿顶圆直径尺寸的测量误差由0.833 mm降至0.544 mm,孔径加键槽尺寸的测量误差由0.133 mm降至0.106 mm,孔径尺寸的测量误差由0.171 mm降至0.148 mm。

Research on accurate reconstruction measurement method of 3D contour point cloud for gears is carried out based on binocular vision measurement system.Aiming at the noise point problem of the initial 3D point cloud of gears,straight pass filtering and voxel raster filtering are used to remove the outliers and redundant points,and moving least squares smoothing plus greedy triangulation reconstruction algorithm is adopted for the reconstruction of gears.The experimental platform of binocular vision gear measurement system is built,and the 3D reconstruction effect and dimension measurement precision of gears are comparatively analyzed through measurement experiments.The results show that the 3D model of the gear reconstructed by moving least squares smoothing plus greedy triangulation has a smoother surface and clearer edges than that of the directly reconstructed gear,and the measurement error of the tooth apex circle diameter dimensions is reduced from 0.833mm to 0.544mm,that of the bore diameter plus keyway dimensions is reduced from 0.133 mm to 0.106 mm,and that of the bore dimensions is reduced from 0.171 mm to 0.148 mm.