Journal of Geo-information Science >
A Hyperspectral Image Classification Algorithm based on the Weighted Exponential Function Model
Received date: 2019-07-18
Request revised date: 2019-11-27
Online published: 2020-10-25
Supported by
National Natural Science Foundation of China(41301479)
Innovative Talents Support Program for Colleges and Universities in Liaoning Province(LR2016061)
Copyright
In recent years, the hyperspectral remote sensing technology has developed rapidly. Hyperspectral images obtained by hyperspectral sensors contain abundant spectral information of ground objects, such that they are good for fine spectral recognition. In hyperspectral image processing, its accurate classification is a solid foundation for the subsequent interpretation tasks. However, the numerous bands in hyperspectral imagery not only provide sufficient characteristics information for classification, but also bring the problem of how to use these characteristics effectively. In this paper, to make full use of the spectral information of hyperspectral images so to achieve accurate classification, a hyperspectral image classification method based on the Weighted Exponential Function (WEF) was proposed that considers the multi-peak characteristics of the spectral response curve of pixels. Firstly, the WEF model was used to build an ideal model of the spectral response curve of pixels, composed of several exponential functions with different weights. Because there are many parameters in the model (including weight, peak position and peak width), it is difficult to solve them. Therefore, the WEF model with fixed peak positions and number of exponential functions was used to model the spectral response curve of all pixels. Then, the parameters of the WEF model were determined according to the least square principle to fit the spectral response curve. Finally, the parameter set was used to replace the spectral measure vector, and the WEF model parameter vector of the pixel was used as its feature. Fuzzy C-means (FCM) algorithm was used for image classification. To validate the feasibility and effectiveness of the proposed method, the classification experiments of Salinas and PaviaU hyperspectral images were conducted by using respectively the proposed method, Principal Component Analysis (PCA) based classification method, Minimum Noise Fraction (MNF) based classification method, and FCM method with the spectral measure vector as the classification feature. The user accuracy, product accuracy, overall accuracy and Kappa coefficient of the results from these classification methods were calculated, and the experimental results were evaluated qualitatively and quantitatively. Compared with other methods, the classification accuracy of the proposed classification method for the Salinas image increased from 51% to 60%, and for the PaviaU image from 43% to 51%. In addition, the proposed classification method reduced the amount of hyperspectral image data while preserving the rich spectral information of hyperspectral images.
LI Yu , LI Yiran , WANG Guanghui , SHI Xue . A Hyperspectral Image Classification Algorithm based on the Weighted Exponential Function Model[J]. Journal of Geo-information Science, 2020 , 22(8) : 1642 -1653 . DOI: 10.12082/dqxxkx.2020.190383
图7 Salinas高光谱图像16类地物光谱曲线拟合Fig. 7 Fitting charts of spectral curves of 16 ground objects in Salinas hyperspectral images |
表1 Salinas各类别光谱曲线拟合精度评价Tab. 1 Evaluation of spectral curve fitting accuracy of Salinas by local matter category |
地物 | C1 | C2 | C3 | C4 | C5 | C6 | C7 | C8 | C9 | C10 | C11 | C12 | C13 | C14 | C15 | C16 |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
光谱角/° | 1.01 | 1.00 | 2.64 | 1.03 | 1.00 | -1.00 | 1.01 | -1.23 | 1.50 | -1.43 | 1.17 | 1.80 | 1.05 | 1.05 | 1.01 | 1.56 |
表2 Salinas的用户精度、产品精度、总体精度、Kappa系数和运行时间Tab. 2 User accuracy, product accuracy, overall accuracy, Kappa coefficient, and elapsed time of Salinas |
分类图像 | 精度/% | C1 | C2 | C3 | C4 | C5 | C6 | C7 | C8 | C9 | C10 | C11 | C12 | C13 | C14 | C15 | C16 |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
未降维 | 用户精度 | 48.50 | 99.65 | 40.01 | 94.66 | 90.49 | 100 | 91.08 | 68.85 | 59.00 | 0 | 16.26 | 5.90 | 43.29 | 90.86 | 49.15 | 0.11 |
产品精度 | 99.75 | 38.30 | 69.03 | 99.21 | 90.96 | 94.47 | 58.20 | 43.89 | 61.45 | 0 | 4.96 | 11.00 | 99.02 | 88.22 | 38.22 | 0.39 | |
总体精度=51.91% Kappa值=0.4755 运行时间:21min | |||||||||||||||||
PCA降维 | 用户精度 | 48.50 | 99.65 | 40.01 | 94.66 | 90.46 | 100 | 91.08 | 68.87 | 58.81 | 0 | 16.26 | 5.71 | 43.29 | 90.86 | 49.14 | 0.11 |
产品精度 | 99.75 | 38.30 | 69.18 | 99.21 | 91.00 | 94.47 | 58.20 | 43.88 | 60.84 | 0 | 4.96 | 10.74 | 99.02 | 88.22 | 38.19 | 0.39 | |
总体精度=51.83% Kappa值=0.4747 运行时间:17min | |||||||||||||||||
MNF降维 | 用户精度 | 58.73 | 99.28 | 0 | 93.39 | 87.27 | 100 | 89.89 | 61.85 | 67.33 | 2.38 | 17.07 | 47.39 | 40.25 | 68.75 | 61.91 | 18.54 |
产品精度 | 99.65 | 59.37 | 0 | 99.35 | 84.47 | 92.12 | 98.10 | 30.85 | 51.10 | 3.69 | 83.90 | 95.23 | 98.47 | 86.36 | 37.52 | 41.06 | |
总体精度=55.08% Kappa值=0.5146 运行时间:18min | |||||||||||||||||
本文算法 | 用户精度 | 59.88 | 98.98 | 41.58 | 94.46 | 94.06 | 100 | 90.41 | 69.05 | 77.74 | 11.56 | 37.71 | 39.61 | 43.25 | 62.49 | 49.22 | 0.13 |
产品精度 | 99.40 | 60.06 | 60.88 | 99.07 | 83.35 | 94.14 | 97.15 | 40.11 | 88.59 | 0.70 | 74.25 | 38.76 | 99.02 | 86.26 | 41.48 | 0.50 | |
总体精度=60.39% Kappa值=0.5682 运行时间:15min |
表3 PaviaU的用户精度、产品精度、总体精度、Kappa系数和运行时间Tab. 3 User accuracy,product accuracy,overall accuracy, Kappa coefficient, and elapsed time of PaviaU |
分类图像 | 精度/% | B1 | B2 | B3 | B4 | B5 | B6 | B7 | B8 | B9 |
---|---|---|---|---|---|---|---|---|---|---|
未降维 | 用户精度 | 79.73 | 87.26 | 0.07 | 44.94 | 25.15 | 20.70 | 0 | 34.62 | 29.97 |
产品精度 | 83.26 | 28.75 | 0.04 | 90.70 | 99.48 | 24.70 | 0 | 42.02 | 99.05 | |
总体精度=43.78% Kappa值=0.3515 运行时间:23 min | ||||||||||
PCA降维 | 用户精度 | 79.65 | 87.81 | 0.06 | 45.06 | 25.36 | 21.95 | 0 | 35.45 | 29.88 |
产品精度 | 83.40 | 28.62 | 0.04 | 90.73 | 99.48 | 26.23 | 0 | 42.53 | 99.05 | |
总体精度=43.97% Kappa值=0.3542 运行时间:19 min | ||||||||||
MNF降维 | 用户精度 | 84.23 | 76.79 | 21.29 | 41.42 | 6.45 | 28.00 | 0 | 44.69 | 99.27 |
产品精度 | 73.52 | 40.93 | 41.07 | 38.35 | 16.88 | 37.28 | 0 | 81.88 | 100 | |
总体精度=48.18% Kappa值=0.3769 运行时间:19 min | ||||||||||
本文算法 | 用户精度 | 79.12 | 78.50 | 9.60 | 48.14 | 89.69 | 29.48 | 0 | 47.73 | 98.64 |
产品精度 | 84.42 | 38.52 | 7.96 | 70.82 | 75.02 | 37.78 | 0 | 86.39 | 99.89 | |
总体精度=51.79% Kappa值=0.4229 运行时间:16 min |
[1] |
张号逵, 李映, 姜晔楠. 深度学习在高光谱图像分类领域的研究现状与展望[J]. 自动化学报, 2018,44(6):961-977.
[
|
[2] |
程希萌, 沈占锋, 邢廷炎, 等. 基mRMR特征优选算法的多光谱遥感影像分类效率精度分析[J]. 地球信息科学学报, 2016,18(6):815-823.
[
|
[3] |
童庆禧, 张兵, 张立福. 中国高光谱遥感的前沿进展[J]. 遥感学报, 2016,20(5):689-707.
[
|
[4] |
蔡悦, 苏红军, 李茜楠. 萤火虫算法优化的高光谱遥感影像极限学习机分类方法[J]. 地球信息科学学报, 2015,17(8):986-994.
[
|
[5] |
陈宏达, 普晗晔, 王斌, 等. 基于图像欧氏距离的高光谱图像流形降维算法[J]. 红外与毫米波学报, 2013,32(5):450-455.
[
|
[6] |
郭新蕾, 张立福, 吴太夏, 等. 成像光谱技术的古画隐藏信息提取[J]. 中国图象图形学报, 2017,22(10):1428-1435.
[
|
[7] |
赵少华, 张峰, 王桥, 等. 高光谱遥感技术在国家环保领域中的应用[J]. 光谱学与光谱分析, 2013,33(12):3343-3348.
[
|
[8] |
王晋年, 李志忠, 张立福, 等. “光谱地壳”计划——探索新一代矿产勘查技术[J]. 地球信息科学学报, 2012,14(3):344-351.
[
|
[9] |
虞佳维, 程志庆, 张劲松, 等. 高光谱信息的农林植被种类区分[J]. 光谱学与光谱分析, 2018,38(12):3890-3896.
[
|
[10] |
汪大明, 肖晨超, 李志忠, 等. 高光谱遥感油气探测技术[J]. 地球科学(中国地质大学学报), 2015,40(8):1301-1309.
[
|
[11] |
任广波, 过杰, 马毅, 等. 海面溢油无人机高光谱遥感检测与厚度估算方法[J]. 海洋学报, 2019,41(5):146-158.
[
|
[12] |
罗素群, 郭宝峰, 沈宏海, 等. 一种定标缺失情况下的高光谱目标识别方法[J]. 火力与指挥控制, 2017,42(6):22-27.
[
|
[13] |
李竺强, 朱瑞飞, 高放, 等. 三维卷积神经网络模型联合条件随机场优化的高光谱遥感影像分类[J]. 光学学报, 2018,38(8):404-413.
[
|
[14] |
李雪轲, 王晋年, 张立福, 等. 面向对象的航空高光谱图像混合分类方法[J]. 地球信息科学学报, 2014,16(6):941-948.
[
|
[15] |
黄鸿, 郑新磊. 高光谱影像空-谱协同嵌入的地物分类算法[J]. 测绘学报, 2016,45(8):964-972.
[
|
[16] |
杜培军, 夏俊士, 薛朝辉, 等. 高光谱遥感影像分类研究进展[J]. 遥感学报, 2016,20(2):236-256.
[
|
[17] |
余铭, 魏立飞, 尹峰, 等. 基于条件随机场的高光谱遥感影像农作物精细分类[J]. 中国农业信息, 2018,30(3):1-9.
[
|
[18] |
王秀和. 利用K均值算法改进后的蚁群优化算法对高光谱图像聚类研究[J]. 科技通报, 2015,31(3):202-206.
[
|
[19] |
纪磊, 张欣, 张丽梅. 基于空谱加权近邻的高光谱图像分类方法[J]. 激光与光电子学进展, 2020,57(6):061013.
[
|
[20] |
臧卓, 林辉, 杨敏华. 利用PCA算法进行乔木树种高光谱数据降维与分类[J]. 测绘科学, 2014,39(2):146-149.
[
|
[21] |
张磊, 邵振峰. 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):114-117,66.
[
|
[22] |
张悦, 官云兰. 聚类与自适应波段选择结合的高光谱图像降维[J]. 遥感信息, 2018,33(2):66-70.
[
|
[23] |
|
[24] |
张莹彤, 肖青, 闻建光, 等. 地物波谱数据库建设进展及应用现状[J]. 遥感学报, 2017,21(1):12-26.
[
|
/
〈 | 〉 |