PCB Component Detection for Hardware Assurance: A Feature Selection-based Approach

Physical inspection in the optical domain is becoming increasingly important to hardware assurance professionals. Several efforts have been undertaken to create a reliable system for Printed Circuit Board (PCB) assurance that combines image processing, computer vision, and machine learning. However, this field remains challenging owing to the need for a significant amount of data as well as emerging problems with the rapid growth of technology. As a result, as demonstrated in our work, feature extraction and feature selection can play a vital role in addressing these problems and developing an interpretable PCB assurance system. Our findings indicate that the color feature, as opposed to the shape and texture features, is the most important for PCB component recognition using semantic data. Furthermore, feature selection can assist alleviate the curse of dimensionality issues and offer a better understanding of how to create an effective PCB component recognition algorithm, which is the goal of our research.