Fuzzy inference filter and morphological operators for short circuits detection in printed circuit board

Surface mounting technology (SMT) is one of the most common techniques for the production of printed circuit board (PCB). During manufacturing process, especially during soldering process, faults, such as short circuits, may happen. For a good efficiency of the production line an automatic system for faults recognition is suitable. In this paper we propose a method based on fuzzy logic inference, for short circuit detection in X-ray images of printed circuit boards. The images are acquired with a digital laminographic technique, in particular with HP 5DX system. We present a fuzzy rule based system that is able to enhance solder joints from the other objects in the laminographic images, such as wire lines and other electronic devices mounted on the bottom side of boards. Fuzzy system results are compared with the outputs of gray scale morphological filter. The fuzzy rule based filtering presents efficiency, robustness and a relative low computational cost. The system presented consists of only two inputs and six rules and is based on the Sugeno-Takagi fuzzy model. Once solder joints enhancement is performed, the image is converted in binary format and an a priori number of binary erosions is performed in order to detect short-circuits.