Research index
天気図を対象とした図形認識実験
谷口, Taniguchi, タニグチ and 6 others
Kyushu University Institutional Repository (QIR) (Kyushu University)9 Oct 1979Open access
天気図を対象とした図形認識実験
谷口, Taniguchi, タニグチ and 6 others · Kyushu University Institutional Repository (QIR) (Kyushu University) · 1979
Abstract
It is one of the most important aspects of artificial intelligence researches to construct a system which has an ability to understand pictures. Generally a picture contains various kinds of information, so it is necessary to select more meaningful information among them. One of good ways is to extract information representable in natural language. The system which we present as an example aims to describe weather charts in natural language. This system consists of three major processes : 1. Extraction of component patterns, 2. Recognition of component patterns, 3. Syntax analysis (Recognition of relations between component patterns). Cognitive process in human beings is considered to use a priori knowledge about the objects. So the system needs to have such knowledge represented in some manner. First we present how a priori knowledge about weather charts aids to recognize them and how it is organized in computer program. Second the fundamental processings such as noise elimination, thinning, template matching and curve following, are presented. Third the effectiveness of these processings are shown through some experimental results. Finally the problems for further research are pointed out.