Development of Visual Inspection System Based on Vector Analysis Technique

The present paper proposes a new concept of image processing method based on vector representation for visual inspection test. The method was applied to detect defects and extract their profiles from digital 2D image of welded rough surface of the structures aiming at the development of automated visual inspection system. The potential availability of the new technique based on vector representation of digital image was investigated through welding defect type recognition problem. It was confirmed that the method gives flexible and useful algorithms for detection and identification of small defect in noisy images. The technique proposed in this paper can be widely applied to various kinds of edge extraction or shape recognition in ambiguous boundary on rough texture.