Abstract: It is required that the weld joint position should be detected and the laser beam focus should be controlled to follow the weld joint accurately during butt joint laser welding. For micro butt joint whose width is less than 0.05 mm, the weldments are magnetized by using an excitation magnetic field. Meanwhile, a magneto optical sensor based on the principle of Faraday magneto effect is applied to acquire the magneto optical images of weld joint. By using the histogram of weld joint magneto optical image as the feature template, and being based on the algorithm of mean shift, a new method for weld detection and seam tracking of laser welding is proposed. Through the study of combining algorithm of particle filter and mean shift, it is found that the weld image target can be searched with the large scale by using particle filter, and the seam tracking can be realized. Also, the feature template of weld joint is updated according to the similarity coefficient of Bhattacharyya to improve the accuracy of seam tracking. Experimental results have showed that the micro weld joint position can be effectively detected from the magneto optical images by using the combining algorithm of mean shift and particle filter, and automatic seam tracking has been realized.

Key words: magneto optical imaging, mean shift, micro weld joint, particle filter, seam tracking