The New Fast Point Feature Histograms Algorithm Based on Adaptive Selection

Wenbin Xie ； Zhen Zhang ； Yuefei Wang ； Yuanyuan Zhang ； Liucun Zhu

《淡江理工學刊》 23卷2期 (2020/06) Pp. 225-232

https://doi.org/10.6180/jase.202006_23(2).0006

摘要

Using Fast Point Feature Histograms feature from point cloud for 3D object recognition or registration, Fast Point Feature Histograms feature descriptor is arbitrarily and inefficiently calculated by subjectively adjusting the neighborhood radius, and the whole process can t be completed automatically. An adaptive neighborhood-selection Fast Point Feature Histograms point cloud feature extraction algorithm has proposed to solve this problem. Firstly, we estimate the point cloud density of many pairs of point clouds. Secondly, compute the neighborhood radius to extract the Fast Point Feature Histograms features for Sample Consensus Initial Alignment registration, and count the radius and the density when the registration performance is optimal, and then the cubic spline interpolation fitting is used to obtain the function expression of the radius and the density. Finally, the Fast Point Feature Histograms feature extraction algorithm has combined with the function to form adaptive neighborhood-selection Fast Point Feature Histograms feature extraction algorithm. The experimental results have shown that the proposed algorithm can adaptively choose the appropriate neighborhood radius according to the density of point cloud, and improve the Fast Point Feature Histograms feature matching performance. At the same time, it is improved the computing speed to a better value, which is of guiding significance.

關鍵字

Fast Point Feature Histograms ； Sample Consensus Initial Alignment ； Point Cloud Density ； Neighborhood Radius ； Adaptive Selection