Defect detection by recursive filters in packaging industry

This paper addresses the task of automating the visual inspection of a strip of plastic by a vision system based on a line scan camera. The defects on the strip are characterized by a local variation in the reflectance properties of the surface to be inspected. There are efficient methods, even with variations in illumination, to highlight the defects. Usually these methods used high-pass filters, to enhance the edges of the defects. In the presence of noise the performance of these filters tends to decrease. Also, before the use of a high-pass filter it is necessary to apply a low-pass filter to smooth the noise in the line image. But in the case of poor contrast defects these methods are inefficient. We propose to take advantage of a new filter which has been developed in our laboratory and which optimizes the edge detection of the defect in any case. The performance of this new filter are demonstrated with real images and with different kind of defects