SPOT5多光谱图像对南沙珊瑚礁信息提取方法的探讨

Research on the Extraction Method of Coral Reef at Spratly Islands Using SPOT5

珊瑚礁在海洋生态系统中具有重要的作用,国外对珊瑚礁遥感信息提取研究较早,但国内起步较晚,目前仍以目视判读为主。使用SPOT5 10 m多光谱数据对我国南沙群岛中弹丸礁和光星礁进行了分类实验,并以QuickBird融合影像判读结果及岛礁调查资料为参考,探讨了SPOT5对珊瑚礁水下信息的辨识能力。首先利用阈值分割方法将水下和水上信息分离;对于水下信息,尝试进行Deglint纠正并引入衍生波段,用4种波段组合方案分别进行最大似然法分类后进行结果比较。结果表明,Deglint纠正能够有效消除水面噪声,从而提高分类精度;近红外波段对水下信息提取有辅助作用,可帮助纠正一些由水深导致的错分;衍生波段替代近红外波段分类效果略差。总之,SPOT5 10 m多光谱数据经过图像纠正后能够有效提取珊瑚岛礁水下成分信息,其分类精度可达80%以上。

Coral reef plays an important role in marine ecosystem.Extracting coral reef information with remote sensing starts late in China and is mainly based on visual interpretation.This paper tries to extract the component information of Swallow Reef and Dallas Reef in Nansha Islands automatically using SPOT5 10 m multispectral data,aiming to explore its identification capability of subsurface information of coral reef.First,the remotely-sensed information was separated into two parts using threshold segmentation method：information above the sea surface and that below the sea surface.After implementing Deglint correction on the latter part of information and introducing a derived band from SPOT5 with it,these images being grouped into 4 different kinds of combinations,were classified using maximum likelihood method.At last,the classification results are compared with QuickBird images,which are regarded as the truth data when the ground data is absent.Results indicate that Deglint correction can reduce the surface noise effectively.NIR band may help avoid the errors in classification caused by the different depth of the sea in some extent.The effect is not as good as expected when the derived band takes place of NIR.In sum,through processes designed in this paper,we could use SPOT5 10 m multispectral data to map components of the coral reef at more than 80% accuracy.