A defect detection scheme for web surface inspection

The goal of this work was to develop an improved defect detection scheme fo r highspeed real-time web surface inspection. This goal was realized by spl itting the task into two independent parts: feature extraction and segmenta tion. Both parts were implemented using efficient algorithms which were imp lemented in hardware that is suitable and fast enough to be included in a w orking web inspection system. The proposed scheme is based on some derived texture features and a new self-organizing map variant, the statistical sel f-organizing map. These techniques offer several improvements over the gray -level thresholding techniques that have been traditionally used in commerc ial web inspection systems.