| 基于纹理和空间特征的中分辨率影像滨海水产养殖用地提取研究 |
| |
| 引用本文： | 王雪峰,冯雪,苏奋振,张宇,王武霞,蒋会平.基于纹理和空间特征的中分辨率影像滨海水产养殖用地提取研究[J].地球信息科学,2018,20(5):694-702. |
| |
| 作者姓名： | 王雪峰 冯雪 苏奋振 张宇 王武霞 蒋会平 |
| |
| 作者单位： | 1. 兰州交通大学, 兰州 7300702. 中国地质大学, 武汉4300743.中国科学院地理科学与资源研究所, 北京 100101 |
| |
| 基金项目： | 国家重点研发技术课题（2016YFC1402902） |
| |
| 摘 要： | 近年来,水产养殖用地分布广泛,但由于其在影像上所表现的复杂性和不均匀性,造成该用地类型提取中的困难,尤其针对中分辨率遥感影像。对此,本文提出了一种基于纹理和空间特征的养殖用地提取方法,该方法主要包括3个步骤：首先,利用纹理熵和归一化差异水体指数NDWI实现水产养殖用地的粗提取;然后,依据相邻地物间的关系实现同类型地物合并;最后,本文构建一种相对宽度作为地物的近似宽度,再次利用NDWI实现水产养殖用地的准确识别。本文以越南万丰湾为研究区域,以Landat-8融合影像（融合后的像元大小为15 m）的目视解译结果为标准,对本文方法与最小距离法分类结果进行比较。实验结果表明,该方法的精度可达91.13%,远高于传统的面向对象方法,并且所提方法的错误率和虚假率分别为0.09%和8.87％,表明了该方法可靠性,因而该方法可为基于中分辨率影像的地物类型提取提供一种有效手段。
|
| 关 键 词： | 水产养殖用地 纹理特征 面向对象 中分辨率影像 |
Extraction of Coastal Aquaculture Land from Medium Resolution Image based on Texture and Spatial Features |
| |
| WANG Xuefeng,FENG Xue,SU Fenzhen,WANG Wuxia,ZHANG Yu,JIANG Huiping.Extraction of Coastal Aquaculture Land from Medium Resolution Image based on Texture and Spatial Features[J].Geo-information Science,2018,20(5):694-702. |
| |
| Authors: | WANG Xuefeng FENG Xue SU Fenzhen WANG Wuxia ZHANG Yu JIANG Huiping |
| |
| Institution: | 1. Lanzhou Jiaotong University,Lanzhou 730070,China2. China University of Geosciences,Wuhan 430074,China3. Institute of Geographic Sciences and Natural Resources Research,Chinese Academy of Sciences, Beijing 100101,China |
| |
| Abstract: | Aquaculture land has become widely distributed in recent years. It's difficult to extract aquaculture land for its complexity and inhomogeneity, especially from the medium spatial resolution satellite images. In this paper, we present an automated method containing three main steps to extract aquaculture land from medium spatial resolution images. First, the texture entropy and Normalized Difference Water Index(NDWI) were used to extracted aquaculture land initially. Then, the interrelations of neighboring objects were utilized to merge objects. Finally, a relative new feature called relative width was proposed, and the NDWI was used again to separate the aquaculture land accurately. This method was applied to Van fong Bay, Vietnam using Landsat-8 images whith pixel size of 15m after image fusion. Manual interpretation was conducted in the same region to support and validate our results of the minimum distance method. The result shows that the precision of the proposed method is 91.13%, which is far higher than the traditional object-oriented method. And the missing rate and false rate of the proposed method are 0.09% and 8.87%, respectively, indicating that the proposed method is reliable. This method provides an accurate and efficient means for fast land use mapping from medium resolution imagery. |
| |
| Keywords: | aquaculture land texture object-oriented middle resolution |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《地球信息科学》浏览原始摘要信息 |
| 点击此处可从《地球信息科学》下载免费的PDF全文 |
|
