Please wait a minute...
浙江大学学报(工学版)
J4  2012, Vol. 46 Issue (7): 1295-1300    DOI: 10.3785/j.issn.1008-973X.2012.07.022
计算机技术﹑电信技术     
空间约束半监督高斯过程下的高光谱图像分类
姚伏天1,2, 钱沄涛1, 李吉明1
1.浙江大学 计算机科学与技术学院,浙江 杭州 310029； 2.中国计量学院 计算机系,浙江 杭州 310018
Semi-supervised learning based Gaussian processes for
hyperspectral image classification
YAO Fu-tian1,2, QIAN Yun-tao1, LI Ji-ming1
1.College of Computer Science and Technology, Zhejiang University, Hangzhou 310029, China;
2. Department of Computer Science and Technology, China Jiliang University, Hangzhou 310018, China
 全文: PDF  HTML
摘要：

针对高光谱遥感图像分类中带标记训练样本较少、导致分类正确率偏低的问题,提出用于高光谱图像分类的空间约束半监督高斯过程方法.由于高光谱图像的特征空间满足流形分布假设,大量未标记样本可以使数据空间变得更加稠密,从而有助于更加准确地刻画局部空间特性,提高分类的精度和普适性.通过对高斯过程模型中的核函数施加空间近邻约束,建立未标记样本与带标记样本之间的空间联系.该半监督高斯过程分类器不仅可以提升高光谱遥感图像的分类性能,而且构造简单,实现方便.实验结果表明,在仅有少量带标记的训练样本情况下,半监督高斯过程分类方法对高光谱图像有较高的分类精度和稳定性.
Abstract:

 A new classification method based on spatial semi-supervised Gaussian processes (SSGP) was proposed to address the problem of low hyperspectral imagery classification performance caused by a small number of labeled training samples. As the feature space of a hyperspectral imagery satisfies the assumption of manifold distribution, a lot of unlabeled samples will make the feature space denser so that the local spatial character can be exploited more precisely and the classification accuracy and generality can be improved. In SSGP, the constraint of spatial neighborhood was imposed into the kernel function of Gaussian process, so the spatial correlations of labeled and unlabeled samples can be embedded in the kernel function. SSGP not only raises the classification performance, but also is easy to build and realize. Experimental results show that SSGP method is very good at classification of hyperspectral images in terms of classification accuracy and stability in the case of small size of labeled training samples．
出版日期: 2012-07-01
:  TP 181  
基金资助:

国家自然科学基金资助项目（60872071,61171151）.
通讯作者: 钱沄涛，男，教授，博导.     E-mail: ytqian@zju.edu.cn
作者简介: 姚伏天(1976- )，男，讲师，从事计算机科学与技术的研究. E-mail: thkdragon@gmail.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:

姚伏天, 钱沄涛, 李吉明. 空间约束半监督高斯过程下的高光谱图像分类[J]. J4, 2012, 46(7): 1295-1300.

YAO Fu-tian, QIAN Yun-tao, LI Ji-ming. Semi-supervised learning based Gaussian processes for
hyperspectral image classification. J4, 2012, 46(7): 1295-1300.

链接本文:

http://www.zjujournals.com/eng/CN/10.3785/j.issn.1008-973X.2012.07.022    或    http://www.zjujournals.com/eng/CN/Y2012/V46/I7/1295

［1］ VATSAVAI R, SHEKHAR S, BURK T. A semisupervised learning method for remote sensing data mining ［C］∥17th IEEE International Conference on Tools with Artificial Intelligence. Hongkong: IEEE, 2005: 5-211．
［2］ GOMEZCHOVA L, CALPE J, CAMPSVALLS G, et al. Semisupervised classification method for hyperspectral remote sensing images ［C］∥International Geoscience and Remote Sensing Symposium. Touluse, France: ［s. n.］,2003: 1776-1778.
［3］ BLUM A, MITCHELL T. Combining labeled and unlabeled data with cotraining ［C］∥Proceedings of the 11th Annual Conference on Computational Learning Theory. Madison, USA: ［s. n.］, 1998: 92-100．
［4］ 赵鹏飞, 钱沄涛, 郑文斌, 等. 基于图像与文本特征的在线生物文献MRI图像库构建［J］. 中国生物医学工程学报, 2010, 29(5):697-703.
ZHAO Pengfei, QIAN Yuntao, ZHENG Wenbin, et al. MRI database construction using image and text features extracted from online biological literature ［J］. Chinese Journal of Biomedical Engineering, 2010, 29(5): 697-703．
［5］ SHAHSHAHANI B, LANDGREBE D. The effect of unlabeled samples in reducing the small sample size problem and mitigating the Hughes phenomenon ［J］. IEEE Transactions on Geoscience and Remote Sensing, 1994, 32(5): 1087-1095．
［6］ SINDHWANI V, CHU W, KEERTHI S. Semisupervised Gaussian process classifiers ［C］ ∥ Proceedings of the 20th International Joint Conference on Artificial Intelligence. Hyderabad, India: ［s. n.］, 2007: 1059-1064．
［7］ MELGANI F, BRUZZONE L. Classification of hyperspectral remote sensing images with support vector machines ［J］. IEEE Transactions on Geoscience and Remote Sensing, 2004, 42(8): 1778-1790．
［8］ 姚伏天, 钱沄涛. 用于高光谱遥感图像分类的空间约束高斯过程方法［J］. 南京大学学报：自然科学版, 2009, 45(5): 665-670．
YAO Futian, QIAN Yuntao. A spatial Gaussian process method for hyperspectral remote sensing imagery classification ［J］. Journal of Nanjing University: Natural Sciences, 2009, 45(5): 665-670．
［9］ CHAPELLE O, SCHOLKOPF B, ZIEN A. Semisupervised learning ［M］. Cambridge: MIT, 2006: 229-231．
［10］ RASMUSSEN C, WILLIAMS I. Gaussian processes for machine learning ［M］. Cambridge: MIT, 2006: 20-26．
［11］ SINDHWANI V, NIYOGI P, BELKIN M. Beyond the point cloud: from transductive to semisupervised learning ［C］∥Proceedings of the 22nd International Conference on Machine Learning. Bonn, Germany: ［s. n.］, 2005: 824-831．
［12］ BELKIN M, NIYOGI P, SINDHWANI V. Manifold regularization: a geometric framework for learning from labeled and unlabeled examples ［J］. Journal of Machine Learning Research, 2006, 7(7): 2339-2434.
[1] 林亦宁, 韦巍, 戴渊明. 半监督Hough Forest跟踪算法[J]. J4, 2013, 47(6): 977-983.
[2] 李侃,黄文雄,黄忠华. 基于支持向量机的多传感器探测目标分类方法[J]. J4, 2013, 47(1): 15-22.
[3] 王洪波, 赵光宙, 齐冬莲, 卢达. 一类支持向量机的快速增量学习方法[J]. J4, 2012, 46(7): 1327-1332.
[4] 艾解清, 高济, 彭艳斌, 郑志军. 基于直推式支持向量机的协商决策模型[J]. J4, 2012, 46(6): 967-973.
[5] 潘俊, 孔繁胜, 王瑞琴. 局部敏感判别直推学习机[J]. J4, 2012, 46(6): 987-994.
[6] 金卓军, 钱徽, 朱淼良. 基于倾向性分析的轨迹评测技术[J]. J4, 2011, 45(10): 1732-1737.
[7] 顾弘, 赵光宙. 广义局部图像距离函数下的图像分类与识别[J]. J4, 2011, 45(4): 596-601.
[8] 罗建宏,陈德钊. 兼顾正确率和差异性的自适应集成算法及应用[J]. J4, 2011, 45(3): 557-562.
[9] 商秀芹, 卢建刚, 孙优贤. 基于遗传规划的铁矿烧结终点2级预测模型[J]. J4, 2010, 44(7): 1266-1269.
[10] 徐磊, 赵光宙, 顾弘. 成对耦合分类器的多球体预处理方法[J]. J4, 2010, 44(2): 237-242.
[11] 欧立勇, 杜树新. 基于多智能体的公共检测资源协调方法[J]. J4, 2010, 44(1): 81-86+123.