Simulation and optimization of imaging for specular surface defect detection system based on structured light field

In defect detection through machine vision, imaging quality significantly impacts subsequent detection accuracy, especially with specular surfaces presenting challenges due to high reflectivity. This paper aims to enhance defect imaging on specular surfaces. We propose a defect detection system utilizing structured lighting and light field cameras to efficiently capture various types of specular defects. We develop an imaging simulation system for it, ensuring high accuracy and efficiency. Furthermore, we devise evaluation metrics to gauge system imaging quality, thereby forming the optimization objective function for system parameters. By employing gradient descent within the feasible parameter domain, we optimize the system parameters to obtain optimal system imaging. Our findings validate the superiority of the optimal solution, with optimization objective function values reflecting defect image classification accuracy across different system parameters.