结合Procrustes分析法和ICP算法的PICP配准算法

发布时间：2018-11-24 07:24

【摘要】：为了解决传统ICP算法存在查找最近迭代点较复杂、单向查找导致较多的错误点对、收敛函数易陷入局部最优状况的问题,提出一种基于Procrustes分析对ICP算法进行改进的PICP算法.首先通过比较三维空间8个方向上的初始变换参数和迭代点对距离值寻找出点云数据的最优初始变换参数;然后采用双向查找最近迭代点机制优化ICP算法,并将查找到的点对构成新的点云数据;最后通过Procrustes分析法对点云数据求解最小二乘函数,从而获得较高的配准精度,完成ICP算法的最优收敛.通过牙齿点云数据以及兔子标准数据的配准测试表明,文中采用的算法能够解决尺度变换和非均匀点云配准问题,且配准结果收敛较快,配准误差较小.和传统ICP算法相比,文中的PICP配准算法具有全局收敛性高、迭代次数少、抗噪能力强的优点.
[Abstract]:In order to solve the problem that the traditional ICP algorithm has the problem that finding the nearest iteration points is more complex, one-way finding leads to more error point pairs, and the convergence function is easy to fall into the local optimal condition, an improved PICP algorithm based on Procrustes analysis for ICP algorithm is proposed. First, the optimal initial transformation parameters of point cloud data are found by comparing the initial transformation parameters in eight directions of 3D space and the distance values of iterative point pairs. Then the ICP algorithm is optimized by using the bidirectional search nearest iterative point mechanism, and the new point cloud data is formed by the point pairs. Finally, the least square function of point cloud data is solved by Procrustes analysis method, so that the registration accuracy is higher and the optimal convergence of ICP algorithm is completed. The registration test of dental point cloud data and rabbit standard data shows that the proposed algorithm can solve the problem of scale transformation and non-uniform point cloud registration, and the convergence of registration results is fast and the registration error is small. Compared with the traditional ICP algorithm, the proposed PICP registration algorithm has the advantages of high global convergence, less iterations and strong anti-noise ability.
【作者单位】：成都信息工程大学电子工程学院;中国气象局大气探测重点开放实验室;
【分类号】：TP391.41

【参考文献】