| 结合地物空间特性的高光谱图像分类方法研究 |
| |
| 引用本文： | 韩玲,张若岚,谢秋昌.结合地物空间特性的高光谱图像分类方法研究[J].测绘科学,2011,36(3):150-151. |
| |
| 作者姓名： | 韩玲 张若岚 谢秋昌 |
| |
| 作者单位： | 长安大学地质工程与测绘学院/西部矿产资源与地质工程教育部重点实验室,西安,710054 |
| |
| 基金项目： | 交通部西部重点科技攻关项目(200416000000); 陕西省自然科学基金项目(2006D10); 教育部博士学科点基金项目(2009020511004) |
| |
| 摘 要： | 以往的高光谱或多光谱图像分类与识别,往往只关注像元光谱维上的特性,其一切特征统计也只在光谱及波段维上展开。但是自然界的复杂性、混合像元问题的存在,仅靠像元的光谱特性是不够的,常会出现"麻点"现象。针对这一问题,本文提出一种结合地物空间特性的高光谱图像分类方法,其分类过程可以分为两个阶段,第一阶段是基于像元光谱特性的图像分类,获得影像分类图;第二阶段是针对第一阶段的分类结果,结合地物空间特性进行空间后分类处理。试验研究结果表明,该方法能够保持地块的连续性和均一性,同时克服了"麻点"现象,大大提高分类的精度。
|
| 关 键 词： | 高光谱遥感 高光谱图像分类 混合像元 地物光谱特性 地物空间特性 |
Classification method of hyperspectral images combined with spatial characteristic of ground objects |
| |
| HAN Ling,ZHANG Ruo-lan,XIE Qiu-chang.Classification method of hyperspectral images combined with spatial characteristic of ground objects[J].Science of Surveying and Mapping,2011,36(3):150-151. |
| |
| Authors: | HAN Ling ZHANG Ruo-lan XIE Qiu-chang |
| |
| Institution: | HAN Ling,ZHANG Ruo-lan,XIE Qiu-chang(College of Geology Engineering and Geomatics of Chang'an University,Key Laboratory of Western China's Mineral Resources and Geoogical Engineering Ministry of Education,Xi'an 710054,China) |
| |
| Abstract: | The traditional image classification and recognition of the hyperspectral or multispectral image often only pay attention to the spectrum dimensional property of the pixels,and all of its characteristic statistics is carried out on the spectrum and wave band dimension.But the nature is complex,and there are problems of the mixed pixel,it's insufficient to only depend on the spectrum characteristic of the pixel,and it often leads to the appearance ofmadian(pocking mark).In connection with this problem,this a... |
| |
| Keywords: | hyperspectral remote sensing hyperspectral image classification mixed pixel spectrum characteristic of ground object spatial character of ground object |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
