Seam tracking based on Kalman particle filter method with colored noises

During the laser vision based seam tracking process, a median filter is utilized to smooth the weld seam laser stripe image. A thinned image is obtained using intensity gravity method, and the seam center position is calculated using the thinned laser stripe image. The measurement noise is assumed to be colored noise, and Kalman filter algorithm with augmented state variables is utilized to generate the proposed distribution of particle filter method. The particle filter method is applied to estimate the optimal seam center position. A seam tracking experiment suggests that the Kalman particle filter algorithm can restrain the impact of colored measurement noise and improve the accuracy of seam tracking.