Automatic inspection of moving surfaces

In many industrial applications, the inspection of the produced product is an important step to guarantee the needed quality. Examples for that is the inspection of copper laminates as a basis for the production of printed circuit boards (PCBs). They consist of a web of fiber glass which is saturated with epoxy resin and covered with a thin copper foil on both sides. In the last steps of production the copper laminates are cut into sheets of e.g. 610mm x 480mm. These sheets are placed on a transportation system which is feeded by a robot system. After that, the material is visually inspected from both sides for defects which could lead to a short or an open connection on the PCB. The sheets with lower quality are then eliminated from the production process by use of a second robot system. In this paper, the measurement technique for automatic surface inspection is presented which delivers information on the photometric properties of the surface (colour and gloss) and 3D-features in a single pass. The technique is based on the principle of photometric stereo. By means of six groups of flashed LEDs six images of the surface are acquired simultaneously. Images representing the different physical properties of the surface with pixel accuracy are calculated by methods of image fusion; for detection and classification of defects information from the property images is combined. Based on the reliable information on the surface, very challenging applications of surface inspection can be solved with high rates for defect detection, good classification results and a very low false alarm rate. Results from the industrial application are also presented.