J4 ›› 2015, Vol. 42 ›› Issue (2): 45-51.doi: 10.3969/j.issn.1001-2400.2015.02.008

• 研究论文 • 上一篇    下一篇

贝叶斯集成框架下的极化SAR图像分类

陈博1;王爽1;焦李成1;刘芳2;毛莎莎1   

  1. (1. 西安电子科技大学 智能感知与图像理解教育部重点实验室,陕西 西安 710071; 2. 西安电子科技大学 计算机学院,陕西 西安 710071)
  • 收稿日期:2014-03-31 修回日期:2014-05-08 出版日期:2015-04-20 发布日期:2015-04-14
  • 通讯作者: 陈博
  • 作者简介:陈博(1985-),女,西安电子科技大学博士研究生,E-mail: chenbo8505@163.com.
  • 基金资助:
    国家重点基础研究发展计划资助项目(2013CB329402);国家自然科学基金资助项目(61271302,61272282,61202176,61271298),国家教育部博士点基金资助项目(20100203120005)

Polarimetric SAR image classification via naive Bayes combination

CHEN Bo1;WANG Shuang1;JIAO Licheng1;LIU Fang2;MAO Shasha1   

  1. (1. Ministry of Education Key Lab. of Intelligent Perception and Image Understanding, Xidian Univ., Xi'an 710071, China; 2. School of Computer Science and Technology, Xidian Univ., Xi'an 710071, China)
  • Received:2014-03-31 Revised:2014-05-08 Online:2015-04-20 Published:2015-04-14
  • Contact: CHEN Bo

摘要: 对于极化合成孔径雷达(SAR)图像,由于雷达角度和地物形状导致属于同一类的数据可能存在较大的差异性. 针对此问题提出了一种基于贝叶斯集成框架的极化SAR图像分类方法. 该算法采用贝叶斯集成,通过学习不同个体而获得的分类面来改善极化SAR图像分类性能. 首先,输入极化SAR图像,并获得其对应的极化SAR数据及特征. 从图像的每一类中任意选择像素点作为图像分类的原始训练样本,并对其进行随机划分获得不同的样本子集. 然后,基于获得的样本子集构造对应极化SAR图像的贝叶斯集成框架. 最后,通过构造的贝叶斯集成框架对极化SAR图像进行分类. 特别在构造贝叶斯集成框架中采用支撑矢量机作为基本的分类器算法.实验结果表明,所提出的算法相比经典的极化SAR分类方法和单个SVM的极化SAR分类方法获得更好的分类性能.

关键词: 极化合成孔径雷达, 图像分类, 贝叶斯集成

Abstract: For PolSAR data, the pixels in the same class may have different appearances because of the topographical slopes and the radar look angle. To improve the image classification performance, a supervised polarimetric synthetic aperture radar image classification method is proposed based on Naive Bayes Combination. In the proposed method, the Naive Bayes Combination is adopted to learn different training samples to get classification surfaces in order to improve the classification results. Firstly, we extract some features and choose some pixels as the original training samples for the classification, and randomly divide the training samples into several training sample subsets. After that, the frame of Naive Bayes combination is obtained based on the training sample subsets. Finally, Naive Bayes Combination gives the final classification results. The support vector machine is used as the basic classifier algorithm in this paper for constructing the Naive Bayes Combination. The experimental results of L-band and C-band data of San Francisco demonstrate the effectiveness and robustness of the proposed method.

Key words: polarimetric synthetic aperture radar (PolSAR), image classification, naive Bayes combination