Search

One of the most important stages in automatic visual inspection for specular surface objects is to generate an image data including features of defects for the inspection. We define the image for inspection as an image data including an intensity change due to surface defects. The type of surface defects on the specular object is classified into local reflectivity losses and local unevennesses. In this paper, we report a study on illuminating, imaging and image processing methods to generate the image for inspection in order to detect the local unevennesses. We considered a situation where a light emitting source displays a sinusoidal intensity pattern and a camera captures the reflection pattern on the specular surface. Numerical computations based on our theoretical analysis shows that the local unevennesses appears as phase gaps of the sinusoidal intensity pattern in the image data captured by the camera. We propose a processing method to convert the phase gaps into the intensity changes in the image for inspection. In addition, we mentioned operational problems to be solved when we apply our methods to an automatic visual inspection system. Furthermore, we discussed a possible method for detecting the local unevennesses with a depth of 1x10-3 mm from the image for inspection.