一种基于数学形态学的遥感影像分割方法  

An segmentation method of remotely sensed imagery based on mathematical morphology

作　　者：纪松[1] 张一鸣[1] 龚辉[1] 

机构地区：[1]信息工程大学测绘学院,河南郑州450052

出　　处：《测绘工程》2007年第4期33-36,40,共5页Engineering of Surveying and Mapping

基　　金：国家高技术研究发展计划(863计划)资助项目(2003AA783060)

摘　　要：介绍形态学测地重建与开闭重建运算,提出基于微分形态学断面的影像分割方法。该方法首先生成原始影像的微分形态学断面影像,然后通过比较得到像点的最大微分断面值以及相对应的特征结构元素,最后以特征结构元素为层次,逐层分割影像。以IKONOS卫星影像为例,该方法能够有效地实现影像的分割,在分割较低反差的影像部分等方面具有优势。Morphological geodesic reconstruction and opening and closing reconstruction operations are introduced, and an image segmentation method based on differential morphological profile is proposed. This method first builds the differential morphological profile image of the original image, and then obtains the pixel＇ s maximum differential profile value and its corresponding feature structure element, and finally, on the feature structure element layers, completes image segmentation process. Experimenting with IKONOS imagery, this method proves to be able to effectively segment images and have great advantages on segmentation images with low contrast areas.

关 键 词：测地重建 开闭重建 微分形态学断面 结构元素 影像分割 

分 类 号：P237[天文地球—摄影测量与遥感] TN911.73[天文地球—测绘科学与技术]