基于聚类和最佳指数的快速高光谱波段选择方法

郭彤; 华文深; 刘恂; 刘晓光

光学技术, 2016, 42 (6): 496, 网络出版: 2016-12-23

基于聚类和最佳指数的快速高光谱波段选择方法

Rapid hyperspectral band selection approach based on clustering and optimal index algorithm

论文大纲

郭彤 ^*华文深刘恂刘晓光

作者单位

军械工程学院电子与光学工程系, 河北石家庄 050003

高光谱波段选择图像分类最佳指数法 K-means聚类 hyperspectral imagery band selection classification optimal index K-means clustering

摘要

最佳指数法是常用的高光谱图像数据波段选择方法, 但存在运算时间过长的问题。运用K-means聚类算法, 对最佳指数方法进行了改进, 提出了聚类最佳指数法, 并进行了一系列伪装目标识别的对比实验。实验结果表明, 与最佳指数法相比, 改进后的方法在保证目标分类精度的前提下, 运算速度提高了数十倍; 与单纯使用K-means聚类运算相比, 不仅运算时间缩短, 而且分类精度有所提高。利用改进算法能够在伪装环境下更加快速有效地识别目标。

Abstract

Optimal index algorithm is a frequently-used hyperspectral band selection approach, but it takes too much computation time. Using the concept of K-means clustering algorithm, the clustering optimal index algorithm is presented to improve the optimal index algorithm and a series of contrast experiments for camouflage target recognition are conducted. The results indicate that, compared with optimal index method, in guarantee the premise of target classification accuracy, computational speed of the improved method increases dozens of times. Meanwhile, compared with K-means clustering algorithm, the operation time is shortened, the classification accuracy is improved. In camouflage environment, the target can be identified more quickly and effectively by the modified algorithm.

参考文献

[1] 刘恂. 光学伪装的高光谱图像探测与评价方法研究［D］. 石家庄: 军械工程学院, 2014.

LIU X. Research of camouflage target detection and evaluation in hyperspectral image［D］. Shijiazhuang: Mechanical Engineering College, 2014.

[2] 田野. 高光谱遥感图像降维方法研究［D］. 哈尔滨: 哈尔滨工程大学, 2008.

TIAN Y. Research on dimensional reduction method of hyperspectral remote sensing images［D］. Harbin: Harbin Engineering university, 2008.

[3] 刘春红,等. 一种新的高光谱遥感图像降维方法［J］. 中国图像图形学报, 2005, 10(2): 218-222.

LIU C H,et al. A new method of hyperspectral remote sensing image dimensional reduction［J］. Journal of Image and Graphics,2005, 10(2): 218-222.

[4] 范伟,等. 减背景的波段选择算法及其在目标探测中的应用［J］. 强激光与粒子束, 2005, 17(6): 828-836.

FAN W,et al. Wavelength band selection algorithm based on background radiance subtraction and its application to target detection［J］. High Power Laser and Particle Beams, 2005, 17(6): 828-836.

[5] 于绍慧,等.基于矩阵模式的高光谱波段选择方法［J］. 光电工程, 2012, 39(6): 71-75.

YU S H, et al. An eifficient hyperspectral band selection method based on matrix mode［J］. Opto-Electronic Engineering, 2012, 39(6): 71-75.

[6] JOSE M B D, et al. Hyperspectral Remote Sensing Data Analysis and Future Challenges［J］. IEEE Transactions on Geoscience and Remote Sensing, 2013, 1(2): 6-36.

[7] 杨金红. 高光谱遥感数据最佳波段选择方法研究［D］. 南京: 南京信息工程大学, 2005.

YANG J H. Optimal band selection methods of hyperspectral remote sensing data［D］. Nanjing: Nanjing University of Information Science and Technology, 2005.

[8] 张磊,等. 改进的OIF和SVM结合的高光谱遥感影像分类［J］. 测绘科学,2014, 39(11): 114-117.

ZHANG L,et al. Hyperspectral remote sensing image classification based on improved OIF and SVM algorithm［J］. Science of Surveying and Mapping, 2014, 39(11): 114-117.

[9] PATRA S,et al. A fast cluster-assumption based active-learning technique for classification of remote sensing images［J］. IEEE Transactions on Geosciences and Remote Sensing, 2011, 49(5): 1617-1626.

[10] 张良培,等. 高光谱遥感［M］. 北京: 测绘出版社, 2011.

ZHANG L P,et al. Hyperspectral remote sensing［M］. Beijing: Surveying and Mapping Press, 2011.

[11] 赵芳,等. 运用资源三号卫星影像数据提取水体信息的研究方法［J］. 测绘通报, 2014,(3): 25-28.

ZHAO F, et al. Study on water body information extraction method based on ZY-3 imagery［J］. Bulletin of Surveying and Mapping, 2014(3): 25-28.

[12] 李荟娆. K-means聚类方法的改进及其应用［D］. 哈尔滨: 东北林业大学, 2014.

LI H R. Improved K-means clustering method and its application［D］. Harbin: Northeast Agricultural University, 2014.

[13] 王志社,等. 基于聚类分割和形态学的可见光与SAR图像配准［J］. 光学学报. 2014, 34(2): 0215002.

WANG Z S, et al. Optical and SAR image registration based on cluster segmentation and mathematical morphology［J］. Acta Optica Sinica, 2014, 34(2): 0215002.

郭彤, 华文深, 刘恂, 刘晓光. 基于聚类和最佳指数的快速高光谱波段选择方法[J]. 光学技术, 2016, 42(6): 496. GUO Tong, HUA Wenshen, LIU Xun, LIU Xiaoguang. Rapid hyperspectral band selection approach based on clustering and optimal index algorithm[J]. Optical Technique, 2016, 42(6): 496.

基于聚类和最佳指数的快速高光谱波段选择方法

关于本站 Cookie 的使用提示

全站搜索

基于聚类和最佳指数的快速高光谱波段选择方法

相关论文

相关资讯

关于本站 Cookie 的使用提示

全站搜索