利用CNN特征和BoWs的三维模型检索算法  

3D model retrieval algorithm based on CNN feature and Bo Ws model

作　　者：董水龙 李海生[1,2] 祝晓斌 蔡强[1,2] 

机构地区：[1]北京工商大学计算机与信息工程学院,北京100048 [2]食品安全大数据技术北京市重点实验室,北京100048

出　　处：《广西大学学报（自然科学版）》2017年第5期1787-1792,共6页Journal of Guangxi University（Natural Science Edition）

基　　金：国家自然科学基金资助项目(61402023);北京市自然科学基金资助项目(4162019);北京市科技计划项目(Z161100001616004)

摘　　要：为提高基于视图的三维模型检索算法的精确度,将卷积神经网络(CNN)特征和词袋模型(Bo Ws)检索思路相结合,提出一种新的基于视图的三维模型检索算法。提取CNN特征,利用Bo Ws的检索思想对模型单一特征进行合并,消除视图间关联。将多个CNN特征融合,提高检索精确度。在ETH-80数据集上实验,结果表明:与目前现有算法相比,本算法检索精确度较高。In order to improve the accuracy algorithm of view-based 3 D model retrieval is of view-based 3D object retrieval algorithm, a new proposed, which combines the features of convoluted neural network and Bag-of-Words（BoWs） model. The CNN features are extracted firstly. Then all features from multiple views of a 3D model are merged into one vector by using the retrieval idea of BoWs to eliminate the relation between views. The multiple CNN features are fused to improve retrieval accuracy. Experimental results on ETH- 80 data set show that the proposed method improves the performance significantly compared with the state-of-the-art approaches.

关 键 词：卷积神经网络特征 词袋模型 特征融合 三维模型检索 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]