An efficient and robust method for detecting light stripe centers in structured light 3D measurement

Abstract The extraction of the center points of light stripes is crucial for line-structured light 3D measurement systems. High-precision center point extraction requires first determining the cross-sectional orientation of the light stripe. However, most researchers treat this as an independent segment of structured light 3D measurement and incur significant costs in calculating the light stripe's normal vector, making it challenging to simultaneously meet the requirements of efficiency, high precision, and robustness. This paper presents a sub-pixel light stripe center point extraction method utilizing light plane calibration information. By projecting the normal of the light plane obtained during the system calibration onto the imaging plane, the direction for center point searching is obtained. Subsequently, sub-pixel expansion is performed to precisely locate the center point of the light stripe. The proposed method optimizes the center detection process without introducing additional measurement stages, reducing computational costs and enhancing interference resistance. The simulation analysis and experiments conducted demonstrate that the proposed method achieves higher accuracy and robustness with minimal time consumption, validating its effectiveness.