点云配准RANSAC算法研究改进

Research and Improvement of Point Cloud Registration RANSAC Algorithm

为了解决机器视觉在复杂环境情况下配准性能不佳的问题,根据被配准目标内部各特征点间的交比不变原理,对RANSAC快速点云配准算法进行了改进。该算法建立了一套被配准目标的模型,通过该模型可以准确计算出目标点云和参考点云之间的仿射矩阵;算法通过对将内点数据间的交比不变量作为配准方法的约束条件,对配准的特征点进行筛选和修正;并将配准过程中参考图像和目标图像配准正确时间和配准准确度作为配准效果优劣的评价标准。最后,将改进后的算法和传统算法进行比较,实验结果显示,改进后的算法不仅在大规模3D点云配准中具有良好的鲁棒性,而且在保证配准精度的前提下提高了配准效率。总体而言,改进后的算法效果明显优于传统算法,在现实应用中具有很高的指导价值,可以大大提高点云配准的配准效率。

In order to solve the problem of poor registration performance of machine vision in complex environment,a fast point cloud registration algorithm based on RANSAC is improved according to the principle of invariant cross ratio among the feature points in the registered object. A model of the registered object is established through which the affine matrix between the target point cloud and the reference point cloud can be calculated accurately;the registration feature points are filtered and modified by taking the cross-ratio invariant between the interior point data as the constraint condition of the registration method and refers to the registration process. The correct registration time and registration accuracy of image and target image are the criteria for evaluating the effectiveness of registration. Finally,the improved algorithm is compared with the traditional algorithm. The experimental results show that the improved algorithm not only has good robustness in large-scale 3 D point cloud registration,but also improves registration efficiency on the premise of ensuring registration accuracy. Overall,the improved algorithm is better than the traditional algorithm and has a high guiding value in practical applications;the registration efficiency of point cloud registration can be greatly improved.