基于知识的遥感图像分类系统

A Knowledge-Based Classifying System for Remote Sensing

本文应用问题求解的黑板模型并利用产生式系统作为基本软件结构,设计并实现了一个基于知识的遥感图像分类系统;对知识获取与表示进行了研究;通过实验论证了系统的可行性.

Based on a blackboard model, a knowledge-based classifying system of images for remote sensing is developed with a production system software. The system developed is composed of an interface, a geographical information acquisition subsystem, pre-processing module, control module, knowledge acquisition module, classifying subsystems, blackboard global database, interpretation module and image displaying module. It is programmed with C language and is easy to modify. The region analysis with this system is based on the fuzzy clustering segmentation and certain defects due to pixel-by-pixel analysis are thus overcome. Relatively stable symbol features can be extracted by using the relative information measure. The feasibility of the system is demonstrated by classifying experiments on TM (Thematic Mapper) images.