Abstract: The current Remote Sensing(RS) image interpretation systems can not meet the given actual demand for tiny objects and the extraction applicability to different objects of interests. This paper designs and implements a new interactive RS image interpretation system scheme based on GrabCut algorithm. An integrative database of all the interested objects is built, which supplies references to their respective Regions Of Interests’(ROI) determination. Different objects are interactively extracted preparing for the following recognition. An experiment takes planes as an example, the experimental results show that the scheme is independent of particular object characteristics, and it provides a good support for the RS image interpretation work.

Key words: Remote Sensing(RS) image interpretation, interactive, GrabCut algorithm

摘要： 当前各种遥感解译系统无法对多种不同目标提供一致的交互解译，且无法满足弱小目标特定的解译需求，针对该问题，基于GrabCut算法设计实现一种交互式的遥感解译识别系统。包括不同目标的特征数据库建立、感兴趣区域定位、交互式目标提取及目标解译识别等主要模块。对飞机目标的解译示例表明，该解译系统不依赖于具体的目标特征，可为遥感图像的解译工作提供良好的支持。

关键词: 遥感图像解译, 交互式, GrabCut算法

CLC Number: 

• TP75