基于H_(Pal)信息熵融合的三维模型检索方法  

H_(Pal) Information Entropy Based Combination Methods for 3D Model Retrieval

作　　者：陈俊英[1] 何波[1] 王羡慧[2] 

机构地区：[1]西安建筑科技大学信息与控制工程学院,西安710055 [2]新疆大学信息科学与工程学院,乌鲁木齐830046

出　　处：《系统仿真学报》2012年第9期1777-1779,1789,共4页Journal of System Simulation

基　　金：陕西省自然科学基础研究计划项目(2012JQ8039);陕西省教育厅科研计划项目资助(11JK1036)

摘　　要：为了弥补单种特征提取算法描述模型特征的不足,提出基于HPal信息熵的三维模型多特征动态融合检索算法。针对给定的查询模型,利用各特征集上的检索结果列表动态计算各特征相对应的HPal信息熵,基于HPal信息熵确定融合权值,然后通过距离加权计算模型之间的相似度。与融合中最好单特征集相比,该方法的检索准确率明显提高,并且实验结果也表明该方法比其他3种流行的算法更为有效地提高了检索效果。In order to improve performance of 3D model retrieval by using multi-feature combination, a dynamic multi-feature combination method of 3D model retrieval based on Hpal information entropy was proposed. Given a query model Heat information entropy contained in feature was computed by using retrieval list of each feature set firstly. Then, combination weights were determined based on Heat information entropy. Finally, the similarity between models was got by dynamically weighted combination of distance. Experimental results show that the proposed method has gained much better retrieval accuracy than the best individual feature involved, and even better retrieval performance than the other three popular algorithms.

关 键 词：三维模型检索 动态融合 多特征 信息熵 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]