›› 2014, Vol. 27 ›› Issue (5): 140-.

• 论文 • 上一篇    下一篇

遗传算法在SAR图像目标鉴别特征选择上的应用

刘轩,王卫红,唐晓斌,李鹏   

  1. (电子科学研究院 综合电子信息研究所,北京 100041)
  • 出版日期:2014-05-15 发布日期:2014-05-14
  • 作者简介:刘轩(1989—),男,硕士研究生。研究方向:SAR图像目标鉴别技术。E-mail:344187975@qq.com

Feature Selection for Target Discrimination in SAR Images Based on Genetic Algorithm

LIU Xuan,WANG Weihong,TANG Xiaobin,LI Peng   

  1. (Institute of Integrated Electronic Information,Electronic Research Academy,Beijing 100041,China)
  • Online:2014-05-15 Published:2014-05-14

摘要:

针对SAR图像鉴别特征选择,在候选特征数较多情况下的应用,文中利用遗传算法提出了一种可靠、准确的鉴别特征选择方法。首先介绍了常用于目标鉴别的19个特征。然后利用遗传算法对上述特征进行了选择。并在评价环节中,对原有方法中适应度函数的权值进行了调整。通过对比实验表明,适应度函数权值的调整,能够加速种群的进化速度,提高了鉴别特征选择的准确性和有效性。

关键词: 合成孔径雷达, 目标鉴别, 特征提取, 特征选择

Abstract:

This paper offers a reliable and accurate method based on Genetic Algorithm for the application of feature selection in SAR images.First,19 common features used in target discrimination are described.Then,the GA is applied to select the best feature series from the 19 features.In the fitness session,weights and thresholds of the original fitness function are adjusted.The experimental results show that adjusting weights and thresholds of the fitness function accelerates evolution and increases the accuracy and effectiveness of feature selection.

Key words: synthetic aperture radar;target discrimination;feature extraction;feature selection

中图分类号: 

  • TN958