Abstract
In recent years, image-processing technology has advanced with the progress in digital signal processing and semiconductor technology, etc.. So, image processing technology has been used in the fields of OA, FA, and EA. However, a domain expert must spend a great amount of time applying the technology to his field, because he must understand how to use the software package.
In order to solve this problem, we developed MMS (Mother Machine System), which generates an opti-mized image processing program by using GA (genetic algorithm). It has been used in many fields, such as in inspection systems for semiconductors, etc.. The conventional MMS needed the input of an outline processing procedure in order to generate a program. To input the outline processing procedure was too difficult for people other than experts in the image processing domain.
In order to reduce the difficulties further, we improved the system so that past cases and genetic informa-tion could be used. This framework is called CBR (Case-Based Reasoning). In this paper, we show that the system applying CBR can generate image processing program more powerful than the conventional system.
