摘要
高空间分辨率遥感影像采用传统基于像元分类方法精度较低,本文通过分析高分辨遥感影像特征,采用面向对象的最近邻监督分类方法对QuickBird影像进行分类研究,首先对影像进行对象分割,然后将分割对象信息、形状特征与及上下文联系等特征构成特征空间进行最近邻监督分类,并与传统的基于像元最近邻分类方法分类进行比较分析,结果表明,本方法能够较好的识别高分辨率地物类型,总精度为92.19%,Kappa系数为0.8835,较好地改善分类效果,适合高分辨遥感影像分类。
The object-oriented approach is applied to classify the QuickBird image in LianYungang city. At first, the image is segmented into object features. The shape and affiliation relation join the feature spaces which are used to classify. And later the classification results of object-oriented approach with the nearest neighbor method of classification are compared. We can get a conclusion that the method of classification proposed in this paper can recognize geo-types much better. And the overall accuracy is 92.19;, the coefficient of Kappa is 0. 8835. The suggested method of classification is suitable for classifying high resolution remote sensed image.
出处
《遥感信息》
CSCD
2007年第6期58-61,93,I0006,共6页
Remote Sensing Information
基金
江苏省测绘科研基金项目(JSCHKY200703)资助
关键词
面向对象
分类
影像分割
最近邻法
QUICKBIRD影像
object-oriented approach
classification
image segmentation
the nearest neighbor method
QuickBird image
