基于相关反馈学习的三维目标检索算法

Journal of Systems Engineering and Electronics ›› 2009, Vol. 31 ›› Issue (7): 1733-1737.

基于相关反馈学习的三维目标检索算法

肖秦琨^1,2, 戴琼海¹, 尔桂花¹, 王洁²

1. 清华大学自动化系, 北京, 100084;
2. 西安工业大学电信学院, 陕西, 西安, 710032

收稿日期:2008-04-03 修回日期:2008-07-18 出版日期:2009-07-20 发布日期:2010-01-03
作者简介:肖秦琨(1974- ),男,副教授,博士后,主要研究方向为动态贝叶斯网络及图像检索.E-mail:xiaoqinkun@tsinghua.edu.cn
基金资助:
国家自然科学基金(60432030;60525111;60772048);中国博士后科研基金(2008043410);陕西省教育厅科研项目基金(07JK277)资助课题

3D object retrieval algorithm based on the relevance feedback learning

XIAO Qin-kun^1,2, DAI Qiong-hai¹, ER Gui-hua¹, WANG Jie²

1. Dept. of Automation, Tsinghua Univ., Beijing 100084, China;
2. School of Electronic Information Engineering, Xi’an Technological Univ., Xi’an 710032, China

Received:2008-04-03 Revised:2008-07-18 Online:2009-07-20 Published:2010-01-03

摘要/Abstract

摘要： 针对三维目标(3D object)检索问题,提出了一种基于混合描述符及多支持向量机融合相关反馈学习的3D目标检索方法。在分析现行3D模型检索方法的基础上,提出了混合描述符H_D及相关反馈学习的总体思路。讨论了H_D框架构建,即在光场图像阵列自适应聚类基础上,分别实现H_D各个子描述符。讨论了基于多支持向量机的融合分类学习机制,并将其用于3D目标检索反馈学习环节,对H_D检索性能及相关反馈学习分类进行了实验分析,结果表明所提出的方法是有效的。

Abstract: A novel content-based retrieval algorithm is proposed for meeting 3D object retrieval.Firstly,based on the analysis of shortcomings of existing view-based 3D model retrieval algorithm,so the concept of lightfield hybrid descriptor(denoted as H_D) is presented and 3D object retrieval frame based on multi-feature relevance feedback learning is developed.Then overall framework H_D,which is made of 3 sub-descriptors,is discussed and how to construct H_D is described.Secondly,the 3D object retrieval algorithm based on multiple support vector machines fusion and the relevance feedback learning is used to improve accuracy and effective of 3D object retrieval.In the end,some experiments are done to analyze retrieval performances based on H_D and relevance feedback learning.The results show that the proposed method is effective.

中图分类号:

TP18

肖秦琨, 戴琼海, 尔桂花, 王洁. 基于相关反馈学习的三维目标检索算法[J]. Journal of Systems Engineering and Electronics, 2009, 31(7): 1733-1737.

XIAO Qin-kun, DAI Qiong-hai, ER Gui-hua, WANG Jie. 3D object retrieval algorithm based on the relevance feedback learning[J]. Journal of Systems Engineering and Electronics, 2009, 31(7): 1733-1737.

参考文献

[1] Del Bimbo A,Pala P.Content-based retrieval of 3D models[J].ACM Trans.on Multimedia Computing,2006.2 (1):20-43.
[2] Shih JL,Lee C H,Wang J T.A new 3D model retrieval approach based on the elevation descriptor[J].Pattern Recognition,2007,40(1):283-295.
[3] Chen D Y,Tian X P,Shen Y T,et al.On visual similarity based 3D model retrieval[J].Computer Graphics Forum,2003,22(3):223-232.
[4] Ansary T F,Daoudi M,Vandeborre J P,et al.A Bayesian 3-D search engine using adaptive views clustering[J].IEEE Trans.on Multimedia,2007,9(1),78-88.
[5] Leifman George,Meir Ron,Tal Ayellet.Semantic-oriented 3D shape retrieval using relevance feedback[J].Visual Computer,2005,21(8):865-875.
[6] Leng Biao,Qin Zheng,Li Li-quan.Support vector machine active learning for 3D model retrieval[J].Journal of Zhejiang University.Science A,2007,8(12).1953-1961.
[7] Cao Liangliang,Liu Jianzhuang,Tang Xiaoou.3D object retrieval using 2D line drawing and graph based relevance feedback[C]//Proc.of the 14th Annual ACM International Conference on Multimedia,2006:105-108.
[8] Chiang Cheng-Chieh,Chan Li-Wei,Hung Yi-Ping,et al.Content based object movie retrieval and relevance feedbacks[J].Eurasip Journal on Advances in Signal Processing,2007.89691.
[9] Chan S C,Ng K T,Gan Z F,et al.The plenoptic video[J].IEEE Trans.on Circuits and Systems for Video Technology,2005,15(2):1650-1659.
[10] Levoy M,Hanrahan P.Light field rendering[J].Computer Graphics,1996:31-42.
[11] Friedman N,Itsik Pe'er,Tal Pupko,et al.A structural EM algorithm for phylogenetic inference[J].Journal of Computational Biology,2002,9:331-353.
[12] Chang H,Yeung D Y.Robust locally linear embedding[J].Pattern Recognition,2006,39(6):1053-1065.
[13] Mavroforakis Michael E,Theodoridis Sergios.A geometric approach to support vector machine (SVM) classification[J].IEEE Trans.on Neural Net-works,2006,17(3):671-682.
[14] Kuhl F P,Giardina C R.Elliptic fourier features of a closed contour[J].Computer Graphics and Image Processing,1982,18:236-258.
[15] Zhou X S,Huang T S.Small sample learning during multimedia retrieval using biasmap[C]//CVPR,2001,1:11-17.
[16] 周建新,高科,李锦涛,等.图像检索中一种有效的SVM相关反馈算法[J].计算机辅助设计与图形学学报,2007,19(4);535-540.
[17] Tieu K.Viola PI Boosting image retrieval[C]//Proc.of Computer Vision Pattern Recognition,Hilton Head Island,South Carolina,2000:228-235.