用二维阶次规正不变矩识别三维目标

RECOGNITION OF 3D OBJECTS WITH ORDER NORMALIZED MOMENT INVARIANTS

论述了用二维投影图像的矩不变量来识别三维目标的方法．该方法有效地降低了目标识别的维数，简化了目标匹配和识别的算法．通过考察阶次规正不变矩ＯＮＭＩ的性能，证明ＯＮＭＩ有效地减小特征量的动态范围并提高了其抗噪性．分类结果表明。

Based on the shift,rotation and scale invariants,this paper presents a method of recognizing 3D objects with its 2D silhouette.The method effectively reduces the dimension of the feature vector and simplifies the matching procedure. Using HMI ,the variance normalized HMI and the order normalized moment invariants as the feature vectors,the paper compares the characteristics of the classification experimental results,it concludes that using the order normalized moment invariants as the feature vector one may abtain better classification result than that of the rest two.