Calibration method based on virtual phase-to-coordinate mapping with linear correction function for structured light system

Abstract

Structured light systems are crucial in fields requiring precise measurements, such as industrial manufacturing, due to their capability for real-time reconstructions. Existing calibration models, primarily based on stereo vision (SV) and pixel-wise approaches, face limitations in accuracy, complexity, and flexibility. These challenges stem from the inability to fully compensate for lens distortions and the errors introduced by physical calibration targets. Our work introduces a novel calibration approach using a virtual phase-to-coordinate mapping with a linear correction function, aiming to enhance accuracy and reduce complexity. This method involves traditional stereo calibration, phase processing, correction with ideal planes, and fitting a pixel-wise linear correction function. By employing virtual samples for phase-coordinate pairs and computing a pixel-wise correction, our methodology overcomes physical and numerical limitations associated with existing models. The results demonstrate superior measurement precision, robustness, and consistency, surpassing conventional stereo and polynomial regression models, both within and beyond the calibrated volume. This approach offers a significant advancement in structured light system calibration, providing a practical solution to existing challenges