针对三维模型检索中D2形状分布算法的改进

Improvement for Retrieval of 3D Models Based on D2 Shape Distribution

如何减少检索时间和提高检索效率是三维模型检索中的2个关键问题。针对Osada等的D2形状分布算法对外形相似而细节不同的模型区分度不高的问题,通过对三维模型任意2个三角面片重心之间距离的计算,得到模型重心与这2个三角面片的重心的距离的比值,利用统计出来的数据分别构建距离分布直方图和距离比值直方图,利用2个直方图进行模型匹配。实验表明,提高了检索性能。

How to reduce the time and increase the efficiency of retrieval are the key problems in 3D model retrieval. The retrieval performance of the Osada＇s D2 shape distribution is not sufficient, failing to distinguish shapes that are quite different. A novel method for 3D models retrieval is proposed .Computing the distance between two barycenters of random triangles, then computing the distance ratio the small distance over the larger distance between the barycenter of the model and the two barycenters which are random chosen., using the statistical data to compute two histograms which are distance histogram and distance ratio histogram. Then the two histograms are used to match the model. Experimental results show the proposed method received a better nerformance.