| 基于光谱相似尺度的支持向量机遥感土地利用分类 |
| |
| 引用本文： | 傅文杰,洪金益,林明森.基于光谱相似尺度的支持向量机遥感土地利用分类[J].遥感技术与应用,2006,21(1):25-30. |
| |
| 作者姓名： | 傅文杰 洪金益 林明森 |
| |
| 作者单位： | ( 1. 中南大学地学与环境工程学院, 湖南 长沙 410083; 2. 莆田学院, 福建 莆田 351100;3. 国家卫星海洋应用中心, 北京 100081) |
| |
| 摘 要： | 提出一种基于光谱相似尺度( spectral similarity scale, SSS ) 的支持向量机( support vector machines, SVM) 遥感土地分类新方法, 该方法选择莆田市作为遥感土地利用分类典型研究区, 利用该区域的Landsat7 ETM 遥感影像结合地面实况调查数据, 从图像上选取少量具有代表性的样本点的光谱作为参考光谱, 利用SSS 方法提取训练样本, 然后应用SVM 算法进行遥感土地利用分类, 并将分类结果与最大似然分类算法( MLC) 相比较, 实验结果表明分类精度上有了很大的提高。
|
| 关 键 词： | 光谱相似尺度 支持向量机 土地利用分类 遥感图像 |
| 文章编号： | 1004-0323（2006）01-0025-06 |
| 收稿时间： | 2005-01-22 |
| 修稿时间： | 2005-11-07 |
A Method of Land Use Classification from Remote Sensing Image Based on Support Vector Machines and Spectral Similarity Scale |
| |
| FU Wen-jie,HONG Jin-yi,LIN Ming-sen.A Method of Land Use Classification from Remote Sensing Image Based on Support Vector Machines and Spectral Similarity Scale[J].Remote Sensing Technology and Application,2006,21(1):25-30. |
| |
| Authors: | FU Wen-jie HONG Jin-yi LIN Ming-sen |
| |
| Affiliation: | ( 1. School of Geoscience and Environmental Engineering, Central South University ,Changsha 410083, China; 2. Putian Univer sity , Putian 351100, China; 3. National Satellite Ocean Application Service, Beijing 100081, China) |
| |
| Abstract: | A new method for land use classification from remote sensing image based on Support Vector Machine(SVM) and Soectoal Similarity Scale(SSS) is presented.The SSS is used to determine spectral similarity by simultaneously measuring the size and shape between two spectrum.SVM,a machine learning algorithm which is based on statistical learning theory,is characteristic in solving limit samples,non-linear and high dimension model recognizing problems,and can be largely used in other arears.Firstly by field investigation,interest region then is set up on the remote sensing image using accurate boundary lines.Select some training sample points and then draw the samples which have been purified to make a sample reference spectrum corresponding with the image's spectrum of each wave band.By drawing some amount of training samples using SSS and constructing classifies using SVM,the land use classification to the whole remote sensing image then can be done.Based on the landsat 7 ETM data and ground true data,the paper takes Putian city as interest region for land use classification using SSS and compares its result with that of MLC.Randomly selecting 200 sample points from each type,it can be seen from the two result images that the precision of this method has reached 89.5%,7.9% higher than that of MLC.The classification speed has also been improved obviously.It has obvious superiority and application prospect. |
| |
| Keywords: | SSS SVM Land use Classification Remote sensing data |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
| 点击此处可从《遥感技术与应用》浏览原始摘要信息 |
|
点击此处可从《遥感技术与应用》下载全文 |
|
