改进的多分辨率点云自动配准算法

Improved Multi-resolution Point Cloud Automatic Registration Algorithm

为解决三维点云配准精度低、速度较慢、易受噪声和外点干扰的问题,提出一种改进的多分辨率点云自动配准算法.算法首先对源点云和目标点云建立KD-tree以加快近邻点的搜索;然后采用基于法向量和特征直方图的配准方法实现粗配准,并对其中的特征提取部分进行了改进,能有效提取特征点,且不会损失大量特征不明显的点云信息.为了进一步提高配准精度,精配准提出一种改进的多分辨率迭代最近点算法,算法提出利用特征点的稠密度计算点云分辨率,同时对关键点采样方法进行了改进.实验结果表明,对于不同规模和含不同程度噪声的点云,此方法在精度、速度、抗噪性方面都得到了改善.

In order to solve the problem of lowaccuracy,slow speed,noise and external point interference of 3 D point cloud registration,an improved multi-resolution point cloud automatic registration algorithm is proposed. The algorithm firstly establishes KD-tree for source point cloud and target point cloud to speed up the search of adjacent points. Then the registration method based on normal vector and feature histogram is used to achieve coarse registration,and the feature extraction part is improved. Feature points can be extracted efficiently without losing a lot of point cloud information with insignificant features. In order to further improve the registration accuracy,an improved multi-resolution iterative closest point algorithm is proposed. The algorithm proposes to calculate the point cloud resolution by using the density of the feature points,and improves the key point sampling method. The experimental results show that this method has improved accuracy,speed and noise resistance for point clouds of different scales and with different degrees of noise.