一个基于多重推理的遥感图片解译专家系统

本文研究了在图象模式识别过程中引入人工智能的知识学习、知识表达和知识推理的理论和技术,设计并实现了一个应用于遥感图片解译的计算机视觉专家系统。 为使系统具有较高的智能化程度,本文提出了一种新的推理方法——多重推理法,它具有处理某些不完全知识和错误知识的能力,并且推理过程所运用的部分知识(规则)是从已知样本中自动归纳学习生成的。

A MULTI-LAYER REASONING METHOD BASED EXPERT SYSTEM FOR REMOTELY SENSED IMAGE INTERPRETATION

In this paper the idea of a knowledge learning, knowledge representation and reasoning method is introduced into image pattern recognition. An computer vision, expert system of remotely sensed image interpretation is designed and realized for practising the above idea.A new reasoning method-a multi-layer reasoning method is worked out. Thismethod has the ability of dealing properly with incomplete and erroneous knowledge. And also a series of special rules used in the multi-layer reasoning procedure can be automatically produced by inductive learning method from some interpreted samples. So the system is more intelligent.