SMD segmentation for automated PCB recycling

Due to the complex structure and the great variability of printed circuit boards (PCBs), an accurate analysis of their material composition and consequently an efficient recycling is currently hardly possible. Facing these problems, we proposed an automated PCB recycling system with a core component for the information retrieval. As a first step towards automated recycling based on a comprehensive analysis of PCBs, the segmentation of surface-mounted devices (SMDs) was presented in this paper. Inspired by the technical background of SMDs, they were divided into two main groups: 1. small devices, such as resistors, capacitors, and 2. integrated circuits (ICs). Then assembly print-based segmentation and color distribution-based segmentation were carried out for the two groups respectively. In comparison to device template-based and solder joint-based approaches, our approach is more flexible and more reliable with respect to the variability of devices and the erosion or soiling of solder joints. As confirmed by the evaluation, satisfying results were obtained. The approach was also evaluated for different imaging conditions so that improved performance can be achieved using appropriate imaging settings.