基于SHOT特征融合的散乱工件点云配准算法

Point Cloud Registration Algorithm for Scattered Workpiece Based on SHOT Feature Fusion

针对随机箱体抓取过程中目标识别和定位问题,提出一种基于SHOT特征融合的点云配准方法.对结构光三维测量获取的点云进行预处理和分割,得到去除噪声点后的多个工件点云数据集;提出基于方向包围盒裁剪的方法,得到去除稀疏边缘点后的工件点云,结合均匀采样算法获取关键点集;通过改进SHOT特征描述子对关键点进行唯一性描述;采用最小方差法查找工件点云的关键点在模板点云中的对应点,根据对应关系求解初始变换矩阵;最后,使用ICP算法进行精确配准,得到工件的精确位姿信息.实验结果表明,将本文算法与基于FPFH特征配准、SHOT特征配准算法进行对比,配准精度分别提高了30. 07%和37. 10%,配准速度分别提高了35. 64%和21. 21%.

Aiming at the problem of target recognition and location in the process of random bin picking,a point cloud registration method based on SHOT feature fusion is proposed. The point cloud obtained from the three-dimensional measurement of structured light is preprocessed and segmented to obtain the point cloud dataset of multiple workpieces. A newmethod based on the oriented bounding box is proposed to get the point cloud of the workpiece after removing the sparse edge points,and to obtain the key point set with the uniform sampling algorithm. The key points is described uniquely by improving the SHOT feature descriptor. The minimum variance method is used to find the corresponding points of the point cloud of the workpiece in the template point cloud,and the initial transformation matrix is solved according to the corresponding relation. Finally,the accurate alignment of the workpiece is obtained by using the ICP algorithm. The experimental results showthat the algorithm is compared with the algorithm based on SHOT feature registration and FPFH,the registration accuracy is increased by 30. 07% and 37. 1% respectively,and the registration speed is increased by 35. 64% and 21. 21% respectively.