Please wait a minute...
浙江大学学报(工学版)
浙江大学学报(工学版)
计算机技术﹑电信技术     
基于超像素分割的空间相关主题模型及场景分类方法
王立军,黄忠朝,赵于前
中南大学 生物医学工程学系,湖南 长沙 410083
New spatial-coherent latent topic model based on super-pixel segmentation and scene classification method
WANG Li-jun, HUANG Zhong-chao, ZHAO Yu-qian
Department of Biomedical Engineering, Central South University, Changsha 410083, China
 全文: PDF(1662 KB)   HTML
摘要：

基于经典的LDA模型,提出新的结合超像素分割技术的空间相关主题模型SP-SLTM及相应的场景分类方法.在建模过程中引入类别约束机制,即给每类场景赋予各自的类主题空间,使模型参数的推导更加简便；在“视觉词包”的生成过程中,对图像区域进行进一步二次超像素分割；提取每个超像素的颜色和纹理特征,形成超像素的混合特征表示.上述方法的优点包括：加上从图像区块所提取的SIFT特征,共得到3种视觉词语,弥补传统方法中采用单一视觉特征描述整幅图像的不足；同一区域内的所有视觉词语共享一个主题,增加视觉词语间的空间相关性.分别将UIUC-Sport数据库的测试结果与CTS-LDA、Spatial-LTM、LDA与pLSA 4种传统方法的测试结果进行比较，结果表明：采用SP-SLTM模型可以比传统方法获得更高的场景分类准确率.
Abstract:

A new spatial-coherent latent topic model using super-pixel segmentation (SP-SLTM) principles and the corresponding method of scene classification were proposed based on typical latent dirichlet allocation (LDA) model. A category constraint mechanism was introduced at the process of modeling topic space. Namely, each scene was restricted to its corresponding class topic simplex, making it easier to deduce  model parameter. Secondly, each image region was further segmented into super-pixels at the generative process of “bags of visual word”. Then, color and texture features of each super-pixel were extracted to form the mixed feature expression of  super-pixels. Advantages of the above methods include two aspects. On the one hand, three kinds of visual words are generated for each image region with the extracted SIFT feature of each image patch, which makes up deficiency caused by  traditional methods using single visual feature. On the other hand, all the visual words in one region belongs to  same topic simplex, enhancing the spatial consistency among the visual words. Compare the results of UIUC-sport database with that of four traditional methods,including  SP-SLTM, CTS-LDA, Spatial-LTM and pLSA,respectively. The results showed that the SP-SLTM model achieved higher classification accuracy than those traditional models.
出版日期: 2015-08-28
:  TP 18/TP 242.62  
基金资助:

中南大学中央高校基本科研业务费专项资金资助项目（2011QNZT013）；国家自然科学基金资助项目（61172184）
通讯作者: 黄忠朝,男,副教授     E-mail: lipse_huang@163.com
作者简介: 王立军（1987-）,男,硕士生,从事智能视频分析和场景理解研究.E-mail:137990239@qq.com
服务　　
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章　　

引用本文:

王立军,黄忠朝,赵于前. 基于超像素分割的空间相关主题模型及场景分类方法[J]. 浙江大学学报(工学版), 10.3785/j.issn.1008-973X.2015.03.002.

WANG Li-jun, HUANG Zhong-chao, ZHAO Yu-qian. New spatial-coherent latent topic model based on super-pixel segmentation and scene classification method. JOURNAL OF ZHEJIANG UNIVERSITY (ENGINEERING SCIENCE), 10.3785/j.issn.1008-973X.2015.03.002.

链接本文:

http://www.zjujournals.com/eng/CN/10.3785/j.issn.1008-973X.2015.03.002    或    http://www.zjujournals.com/eng/CN/Y2015/V49/I3/402

［1］ SZUMMER M, PICARD R W. Indoor-outdoor image classification ［C］. IEEE International Workshop on Content-based Access of Image and Video Database, Bombay:IEEE,1998: 42-50．
［2］ VAILAYA A, FIGUEIREDO M A T, JAIN A K, et al. Content-based hierachical classification of vacation images ［C］∥IEEE International Conference on Multimedia Computing and Systems, Florence:IEEE,1999,1:518-523．
［3］ VALIAYA A, FIGUEIRDO M A T, JAIN A K, et al. Image classification for content-based indexing ［J］. IEEE Transactions on Image Processing, 2001, 10(1)：117-130.
［4］ RUSSAKOVSKY O, LI F F, LIN Y Q, et al. Object-centric spatial pooling for image classification ［C］∥Computer Vision-ECCV, 2012. Berlin Heidelberg:Springer, 2012:115．
［5］ANGELOVA A, ZHU S H. Efficient object detection and segmentation for fine-grained recognition ［C］∥ IEEE Conference on Computer Vision and Pattern Recognition.Portland: IEEE, 2013:811-818．
［6］ LI T, MEI T, HUA X S. Contextual bag-of-words for visual categorization ［J］. IEEE Transactions on Circuits and Systems for Video Technology,2011,21(4)：381-392．
［7］ FERNANDO B, FROMONT E, Muselet D, et al. Supervised learning of Gaussian mixture models for visual vocabulary generation ［J］. Pattern Recognition,2012,45(2)：897-907．
［8］ PARIZI S N, OBERLIN J G, FELZENSZWALB P F. Reconfigurable models for scene recognition ［C］∥IEEE Conference on Computer Vision and Pattern Recognition,Providence: IEEE, 2012:2775-2782．
［9］ NIU Z X, HUA G, GAO X B, et al. Context aware topic model for scene recognition ［C］∥IEEE Conference on Computer Vision and Pattern Recognition, Providence:IEEE,2012: 2743-2750．
［10］ HOFMANN T. Probabilistic latent semantic analysis ［C］∥Proceedings of the Fifteenth Conference on Uncertainty in Artificial Intelligence. Morgan Kaufmann Publishers Incorporated,1999:289-296．
［11］ BLEI D M, NG A, JORDAN M I. Latent dirichlet allocation ［J］. Journal of Machine Learning Research, 2003,3:993-1022．
［12］ BOSCH A, ZISSERMAN A, MUNOZ X. Scene classification via pLSA ［C］. Computer Vision-ECCV 2006. Berlin Heidelberg:Springer,2006:517-530．
［13］ BOSCH A, MUNOZ X, OLIVER A, et al. Object and scene classification: what does a supervised approach provide us？ ［C］∥IEEE International Conference on Pattern Recognition,New York:IEEE,2006,1:773-777．
［14］ QUELHAS P, MONAY F. A thousand words in a scene ［J］. IEEE Transactions on Pattern Analysis and Machine Intelligenc,2007,29(9)：1575-1589．
［15］ ELANGO P K, JAYARAMAN K. Clustering images using the latent dirichlet allocation model ［EB/OL］. University of Wisconsin,2005. http:∥pages.cs.wisc．edu/～dyer/cs766/projects/elango05.pdf
［16］ 唐颖军, 须德, 解文杰, 等. 一种基于类主题空间的图像场景分类方法［J］. 中国图象图形学报,2010,15(7)：1067-1073．
TANG Ying-jun, XU De, XIE Wen-jie， et al. A novel image scene classification method based on category topic simplex ［J］. Journal of Image and Graphics,2010,15(7)：1067-1073．
［17］ CHAI Y N, LEMPITSKY V, ZISSERMAN A. Bicos: A bi-level co-segmentation method for image classification ［C］∥IEEE International Conference on Computer Vision,Barcelona:IEEE,2011:2579-2586.
［18］ CAO L L, LI F F. Spatially coherent latent topic model for concurrent segmentation and classification of objects and scenes ［C］∥IEEE 11th International Conference on Computer Vision.Rio de Janeiro:IEEE,2007: 18．
［19］ DU L, REN L, DAVID B D. A bayesian model for simultaneous image clustering, annotation and object segmentation ［C］∥Advances in Neural Information Processing Systems,Whistler:NIPS,2009:486494．
［20］ LOWE D G. Distinctive Image Features from Scale-Invariant Keypoints ［J］. International Journal of Computer Vision,2004, 60(2)：91-110．
［21］ BLEI D M, JORDAN M I. Variational Inference for Dirichlet Process Mixtures ［J］. Bayesian analysis,2006,1(1)：121-143．
［22］ MORI G. Guiding model search using segmentation ［C］∥IEEE International Conference on Computer Vision. Beijing:IEEE,2005(2)：1417-1423.
［23］ LI L J, SOCHER R, LI F F. Towards total scene understaning: classification, annotation and segmentation in an automatic framework ［C］∥IEEE Conference on Computer Vision and Pattern Recognition.Miami:IEEE,2009: 2036-2043．
［24］ WEIJER J V D, SCHMID C. Coloring local feature extraction ［M］∥ Computer Vision-ECCV2006. Berlin Heidelberg:Springer,2006:334-348．
［25］ VARMA M, ZISSERMAN A. Classifying images of materials: Achieving viewpoint and illumination independence ［M］∥Computer Vision-ECCV2002, Berlin Heidelberg:Springer,2002:255-271.
No related articles found!