Abstract
In a fine laser cutting, the surface quality is affected by the spatter removed from the kerf. Spatter characteristics depend mainly on the laser irradiation conditions and the flow properties of the assist gas. Moreover, these flow properties are mainly affected by the pressure of gas and the choice of nozzle. However, the nozzle optimization requires measurements of material removal in the laser cutting, which are difficult to obtain due to the relatively short time window and small physical size of the cutting event. A novel computer vision method was proposed to measure the velocity distribution of spatter in the laser cutting event. Two high speed video cameras are used to obtain different views of the cutting zone. Particles in the videos are detected using a feature detector and tracked using a sequential Monte-Carlo method. The 3D trajectories of particles are reconstructed from the 2D trajectories with an unscented Kalman filter. The performance of the proposed method was discussed using computer-generated videos with synthetic ground truth, and the proposed method was applied to measure the spatter velocity distributions of actual laser cutting events. The proposed system is able to obtain measurements at a wide range of cutting parameters.
