A neural network approach to machine vision systems for automated industrial inspection

A knowledge-based machine vision system for industrial web inspection that is applicable to a broad spectrum of different inspection tasks is proposed. The main function of the machine vision system is to detect undesirable defects that can appear on the surface of the material being inspected. A neural network is used within the blackboard framework for a labeling verification step of the high-level recognition module of this system. As an application, this system has been applied to a lumber inspection problem. It has been successfully tested on a number of boards from several different species. A performance comparison of the neural network with the k-nearest neighbor classifier is also presented.<
>