滩涂水边线Landsat-5影像提取方法

快速获取遥感影像水边线时空变化信息具有重要意义，滩涂水边线提取一直是遥感技术应用的难点问题。水边线在遥感影像上具有独特的空间关系与光谱特征。本文综合使用颜色模型变换法、信息熵计算法、最大类间方差法及边缘检测方法。以长江口崇明东滩为研究区，研究了Landsat-5卫星影像海陆对比度增强及不同尺度下的边缘提取，重点给出了基于热红外波段的水边线空间特征与光谱特征的计算方法，在面向对象技术框架下提出了一种顾及空间关系和光谱特征的遥感影像水边线快速提取方法。实验结果表明:(1)基于最大类间方差法的局部阈值分割法能够自动提取band 6的水边线，水边线连续、完整，空间信息丰富；(2)综合使用最佳指数法、离散度方法及颜色模型变换方法，能够有效增强海陆对比度，基于最大类间方差法的局部自适应Canny算子能够自动检测出增强后遥感影像高精度边缘；(3)利用水边线的空间关系和光谱特征，能够由计算机自动实现水边线的识别与连接工作；(4)本文提出的水边线提取方法速度快、自动化程度高，分别继承了阈值分割法的连续性强的优点和Canny算子定位精度高、细节呈现能力强的优势。研究结果对于海岸带动态变化、陆海相互作用机制、海岸带资源保护与开发及近海工程管理等研究具有重要的参考价值。

Landsat-5 image extraction method for tidal flat waterline:Take the Chongming Dongtan,Changjiang River Estuary as an example

It is of great significance to quickly acquire spatiotemporal change in the information of waterline of remote sensing image.The extraction of the waterline of tidal flat on the remote sensing image has always been a difficult problem in the application of remote sensing technology.There are unique spatial relationships and spectral characteristics on the remote sensing image of waterline.The research area is the Chongming Dongtan of the Changjiang River Estuary.By integrating methods of color model transformation,information entropy calculation,maximum variance and edge detection,we explored how to enhance the contrast of land and sea on the Landsat-5 satellite image,and the edge extraction at different scales was studied.The calculation method of the spatial and spectral characteristics of the waterline using the thermal infrared band was given.A fast extraction method of waterline of sensory image taking the spatial relationship and spectral characteristics into account under the framework of object-oriented technology was proposed.Results show that:(1)The local threshold segmentation method based on the maximum between-class variance method can automatically extract the waterline of band 6.The waterline is continuous,complete,and rich in spatial information.(2)The combination of the optimum index factor method,the dispersion method and the color model transformation method can effectively enhance the contrast between land and sea.The local adaptive Canny operator based on the maximum between-class variance method can automatically detect the high precision edge of the enhanced remote sensing image.(3)Using the spatial relationship and spectral characteristic of waterline,the computer can recognize and connect waterline automatically.(4)The waterline extraction method proposed in this paper is fast and automated,inheriting strong continuity of the threshold segmentation method and high positioning accuracy and strong ability to present details of Canny operator.The results have significant value for researches on the dynamic changes in the coastal zone,the mechanism of land-sea interaction,the protection and development of coastal zone resources,and offshore engineering management.