以高分一号卫星遥感影像为数据源,引入形态学算法,研究采用面向对象的影像分类方法进行城市建筑物提取的关键技术.研究方法结合影像分割与基于知识规则的影像分类技术,首先采用基于形态学开闭重建的分水岭分割算法对高分影像进行分割,其次采用基于知识规则的SVM分类方法对影像进行分类,达到提取建筑物的目的.结果显示,3个研究区建筑物提取的kappa系数分别为0.85、0.66和0.65,利用基于知识规则的面向对象分类方法对高分辨率遥感影像中建筑物的提取效果较好,能够完整、准确地提取出建筑物外形信息,具有较高的应用与推广价值.%Based on the high resolution satellite-1 remote sensing image,a morphological algorithm is introduced to study on the key technologies of urban building extraction using object-oriented image classification method.To extract the building,this paper combined with the method of image segmentation and classification of knowledge rules which based on image technology.It firstly used watershed segmentation algorithm based the morphological opening and closing reconstructing rules in segmentation of high resolution image,then classified the image by using the SVM classification method based on knowledge rules.The results show that the kappa coefficients of building 3 research area extraction were 0.85,0.66 and 0.65,it indicates that using object oriented classification based on knowledge rules on extraction of buildings in high resolution remote sensing image has higher application value,as it can fully and accurately extract the outline of building information.
展开▼
