基于增强型点对特征的三维目标识别方法

Three-Dimensional Object Recognition Based on Enhanced Point Pair Features

针对基于原始点对特征的三维目标识别算法中存在的内存浪费、效率不高的问题,提出了一种基于增强型点对特征的三维目标识别算法。通过在原始点对特征的第4个分量上乘以一个符号函数,得到了一种区分性更强的点对特征,消除了原始点对特征存在的二义性。考虑到待识别目标三维模型存在的自遮挡,利用点对之间的视点可见性约束,剔除了目标三维模型哈希表中存在的大量冗余点对,节省了内存开销并提高了三维目标识别算法的识别准确率和效率。在开放数据集和实际采集的数据集上的实验结果表明,与基于原始点对特征的算法相比,所提三维目标识别算法在识别准确率和效率上都有一定程度的提升。

Aiming at the problems of memory waste and low efficiency in three-dimensional(3 D)object recognition algorithm based on original point pair feature(PPF),a 3 D object recognition algorithm based on enhanced point pair feature(EPPF)is proposed.By multiplying the fourth component of the original PPF with a sign function,a more distinguishing PPF is obtained,which eliminates the ambiguity of the original PPF.Considering the self-occlusion of the 3 D model of the target to be identified,the large numbers of redundant point pairs existing in the target 3 D model hash table are eliminated by means of the viewpoint visibility constraint between the point pairs,which reduces the memory overhead and improves the accuracy and efficiency of the 3 D object recognition algorithm.The experimental results on the open dataset and the actual collected dataset show that the proposed 3 D object recognition algorithm can improve recognition accuracy and recognition efficiency.