-
文中搭建系统如图6所示,标定实验中使用精密加工的304不锈钢标定筒,内径155 mm,外径175 mm,加工精度为0.02 mm,内表面粗糙度Ra为0.4 μm。
-
为验证文中算法,使用标定筒对检测系统位姿进行标定时。标定步骤如图7所示。首先以固定步长旋转深孔类零件测量系统,对固定内径的直圆筒进行三维重构,由点云采集系统获取不同角度测量区域在CCF中的单视角点云;然后,对CCF中的点云进行圆柱拟合,求取CCF中z轴与柱面交点在柱面中心轴线的投影点作为CyCFi的原点,将点云通过坐标变换转换到CyCFi中;然后,结合逆向重建模型建立CyCFi→CyCF中的坐标转换关系,并通过CCF点云坐标和CyCF中的点云坐标关系求解系统位姿参数,在求解过程中不断调整逆向重建参数,求取损失函数接近最小值收敛时的系统位姿参数。最后,在实际测量中通过标定的系统位姿参数对不同位置的测量点云进行点云拼接,完成深孔的三维重建。
在标定实验中,为提高重建误差模型参数的拟合精度,尽量密集得采集深孔内表面点云,以8º为步长采集标定筒内表面的点云数据,增加重叠区域。按照上述步骤进行系统位姿参数标定实验,标定结果如表1所示。
t1(θ) t2(θ) t3(θ) a1 b1 c1 a2 b2 c2 a3 b3 c3 0.0014 0.0250 0.0062 0.9996 0.0078 0.0024 −0.0078 0.9993 0 t4(θ) t5(θ) t6(θ) a4 b4 c4 a5 b5 c5 a6 b6 c6 1.2186 −79.1166 −0.0280 0.0386 −0.0038 −0.0013 0.0078 −0.9986 0 t7(θ) t8(θ) t9(θ) a7 b7 c7 a8 b8 c8 a9 b9 c9 0.9997 0.0079 0.0025 −79.1178 −1.2321 −0.1902 0 0 0.9986 t10(θ) t11(θ) t12(θ) a10 b10 c10 a11 b11 c11 a12 b12 c12 −0.0145 −0.0020 −0.0029 0 0.0020 −0.0318 −4.7672 −5.9891 4.9857 Table 1. The calibration results of the system pose parameters
-
为评估文中点云拼接方法的性能,使用表1中的标定数据对标定筒内表面进行独立的测量实验,在精度评估实验中,每10º采集标定筒其他区域的内表面点云数据进行拼接,其拼接结果及细节如图8 (a)、(b)所示。将两片点云重叠区域到全局坐标系z轴的径向距离差值作为拼接误差,对点云拼接的重叠区域拼接误差计算结果如图8 (c)所示。由图8 (c)可知,在155 mm直径的深孔类零件测量条件下,点云拼接的多数重叠区域拼接误差要低于0.05 mm,最大误差在0.08 mm以下,说明文中提出算法有效,拼接精度满足实际要求。将拼接的点云结果向xoy平面投影,拟合平面圆周求取测量直圆筒的直径为154.957 mm,能够对深孔内表面的关键参数提取的精度要求。参考文献[11]测量圆柱内径的点云拼接精度为0. 12 mm,参考文献[15]在测量拼接60 mm直径圆柱点云时重叠误差在0.05 mm以下,文中提出的拼接方法精度与之相比均有明显优势。
-
对实际零件进行实验验证,使用深孔类零件内表面检测系统对图9 (a)所示带有凹槽的深孔类零件进行测量,在实验过程中同样以10º为步长旋转采集点云数据。对凹槽部分进行点云拼接实验,重建结果及细节如图 9(b)所示,拼接重叠处过渡光滑,拼接精度满足需求。
Research on point cloud splicing for inner surface inspection of deep hole
doi: 10.3788/IRLA20210210
- Received Date: 2021-03-31
- Rev Recd Date: 2021-07-21
- Publish Date: 2021-12-31
-
Key words:
- optical measurement /
- three-dimensional reconstruction /
- coordinate transformation /
- deep hole inspection
Abstract: The single-view point cloud obtained by the deep hole inner surface inspection system cannot reflect the full picture of the inner surface of the deep hole part. In order to realize the three-dimensional reconstruction of the complete surface shape of the deep hole inner surface, a pose calibration technology for the deep hole inner surface inspection system was proposed, which was used to provide the initial value of the point cloud stitching for the reconstruction of the inner surface of the deep hole. Firstly, the principle of deep hole inner surface detection technology was introduced, the coordinate transformation model of deep hole parts measurement point cloud was analyzed and established, the system pose parameters that need to be calibrated in the coordinate transformation model were determined; Then, the measurement trajectory of the system was described by the rotation axis angle, and the point cloud was reconstructed inversely; Then the measurement trajectory model was optimized with the solution error of the system parameter as the loss function, and the system pose parameter calibration was realized; Finally, point cloud splicing experiments of the straight cylinder and the concave groove show that the method was suitable for the detection of the inner surface of the deep hole and easy to operate, the splicing error was not more than 0.08 mm. Compared with other methods, it has certain advantages.