基于稀疏编码和多核学习的图像分类算法

被引:6
作者
程东阳
蒋兴浩
孙锬锋
机构
[1] 上海交通大学电子信息与电气工程学院
关键词
稀疏编码; 多核学习; 特征融合;
D O I
10.16183/j.cnki.jsjtu.2012.11.016
中图分类号
TP391.41 [];
学科分类号
080203 ;
摘要
提出了一种基于稀疏编码和多核学习的图像分类算法.首先从图像中提取Dense-SIFT(Dense Scale-Invariant Feature Transform)和Dense-SURF(Dense Speeded Up Robust Feature)2种特征,使用稀疏编码对特征点进行处理,得到一系列高维向量,然后对这些高维向量应用max-pooling算法,将图像表示成单个向量.最后,使用改进的多核学习方法对这些向量进行分类,对于不同的特征,使用不同核的组合以达到最好的分类效果.实验结果表明,该算法作为词袋(BoW)模型的改进,能够提高分类准确率.
引用
收藏
页码:1789 / 1793
页数:5
相关论文
共 8 条
  • [1] Multiple kernel learning, conic dualityand the SMO algorithm. Bach F. R,Lanckriet G. R. G,Jordan M. I. Proceedings of the twenty first international conference onMachine learning . 2004
  • [2] Improved Bags-of-Words Algorithm for Scene Recognition. Jiang Hao,Xu Jie. Proc of IEEE International Confer-ence on Signal Processing Systems . 2010
  • [3] Linear Spatial Pyramid Matching Using Sparse Coding for Image Classification. Yang Jianchao,Yu Kai,Gong Yihonget al. Proc.of IEEE Computer Society Conference on Computer Vision and Pattern Recognition . 2009
  • [4] Beyond bags of features:Spatial pyramid matching for recognizing natural scene categories. LAZEBNIK S,SCHMID C,PONCE J. Proc.of IEEE Int.Conf.on Computer Vision and Pattem Recognition (CVPR’’06) . 2006
  • [5] Least angle regression. Efron B,Hastie T,Johnstone I,et al. The Annals of Statistics . 2004
  • [6] Image classification by non-negative sparse coding,low-rank and sparse decompo-sition. Zhang C,Liu J,Tian Q. Proc IEEE Comput Soc Conf Comput Vi-sion Pattern Recognit (CVPR) . 2011
  • [7] Online Learning forMatrix Factorization and Sparse Coding. Mairal J,Bach F,Ponce J,et al. Journal ofMachine Learning Research . 2010
  • [8] Object recognition withfeatures inspired by visual cortex. Serre T,Wolf L,Poggio T. Proc IEEEComput Soc Conf Comput Vision Pattern Recognit (CVPR) . 2005