This article presents the design and implementation of a system for laser automatic welding bead offset tracking and identification during laser processing. The system consists of a structured light sensor, a horizontal offset recognition algorithm and a real-time data smoothing algorithm. It can automatically extracts weld bead feature point data of various shapes in real time under different working conditions. By calculating and processing the feature point data, the weld bead is obtained. The real-time offset data is filtered by a real-time smoothing algorithm and then output to the laser automatic welding device, ultimately achieving automatic control of horizontal tracking. In actual on-site welding tests, the system can collect weld bead offset data in real time and accurately guide the welding laser output head to horizontally track the weld bead, laying the foundation for the realization of automated laser welding.