基于FPFH的点云特征点提取算法

FEATURE POINT EXTRACTION ALGORITHM OF POINT CLOUD BASED ON FPFH

提出一种基于快速点特征直方图(FPFH)的点云特征点提取方法。将FPFH引入到点云模型特征点提取技术中,建立FPFH各子区间均值参数,并将之与点的FPFH各区间的值进行对比,提取初始特征点;利用基于特征点邻近点夹角的方法对脊谷特征点进行优化;采用基于投影点夹角的方法对点云轮廓特征进行增强。实验结果表明,该方法能够有效提取出点云特征点,并能较好地表达模型的细节特征和整体轮廓特征。

We propose a feature point extraction method based on fast point feature histogram(FPFH).FPFH was introduced into the feature point extraction technology of the point cloud model,and the mean parameter of each sub-interval of FPFH was established.Compared with the FPFH values of each point,the initial feature points were extracted.Then,we used the method based on the angle between the adjacent points to optimize the ridge-valley feature points.Finally,the contour features of point cloud were enhanced based on the angle between the projection points.The experimental results show that our method can effectively extract the feature points of the point cloud,and can better express the detailed features and overall contour features of the models.