Laser Beam Centroid Detection for Automatic Spatial Filtering: A Comparative Analysis of Machine Vision Algorithms
摘要
Achieving optimal beam quality in applications ranging from high-power lasers to interferometric metrology fundamentally relies on precise spatial filtering. While spatial filters effectively remove high-frequency noise and improve beam profiles, their performance is critically dependent on accurate alignment - specifically, the precise centering of the focused laser beam onto the pinhole aperture. Despite decades of research and development in optical filtering, a robust, automated solution for multiple wavelength spatial filter alignment remains an open challenge. This paper focuses on a critical sub-problem within that challenge: the accurate and reliable determination of the diffraction pattern’s center, a pre-requisite for any automated alignment system. Was observed that the FBM and CCL are faster that the superpixel methods, but superpixel-based methods were more accurate and precise when its compared to the true centroid, achieving minimum errors of one pixel. An hybrid approach algorithm could be a better solution instead a single one, providing the faster execution time and the consistent accuracy.