A LOW-COST VISION SYSTEM COMBINING CONVENTIONAL AND ARTIFICIAL-INTELLIGENCE TECHNIQUES FOR COMPLEX IMAGE INSPECTION AND VERIFICATION

A hybrid image processing technique is described which combines tradit ional feature extraction and windowing with a new artificial intellige nce technique based on weightless neural networks. This technique is a unique way of compressing and then analysing complex data. Described here is its application to the measuring of light scattered for machin e surfaces as well as the inspection of complex and highly coloured im ages. These examples illustrate the distinctive nature of the system. It is called WISE (Windowed Image Shaped Extraction) because it descri bes the stages of image processing.