一种基于零阶球谐函数的点云识别方法

A point cloud recognition method based on zero order spherical harmonic function

本文方法运用基于零阶球谐函数来解决三维点云的识别问题.其核心在于利用最小二乘法将模型的原始坐标拟合成球谐函数正交基的扩展系数,并且将标量形式的计算过程改为以三个坐标轴方向分别推导的向量形式,目的在于使三维点云识别的过程不受原模型几何形状结构的影响.此识别方法具备平移旋转不变的特性,并且在计算过程中不需要预先对三维点云模型进行对齐操作,进一步缩短了识别时间,提高了运算效率.为解决三维点云识别的实时性问题提供了新方法.

This paper focuses on the recognition of 3D point clouds based on spherical harmonic function.In this method,the original coordinates of the model are fitted to the expansion coefficients of the orthogonal basis of the spherical harmonic function by the least square method,and the scalar derivation is changed to the vector derivation in the direction of three coordinate axes respectively.The purpose of this method is to avoid the geometry of the original model.This recognition method has the characteristics of translation and rotation invariance,and don’t need to align the three-dimensional point cloud model in advance during the calculation process,which further shortens the recognition time and improves the operation efficiency.It provides a new method to solve the real-time problem of 3D point cloud recognition.