A reference center extraction method for SG-II-UP CSF alignment control

Aimed that the image applied in CSF(Cavity Spatial Filter)auto-alignment control of SG-II-UP(Shenguang II Laser Upgraded Device) is blurred with uneven light radiation and has many scattered reference points, a novel algorithm of reference center extraction is proposed in this paper, which is based on position estimation and gradient extremum search. Firstly, all the reference points are determined as clear or obscure by simple threshold method. Then the clear reference points are applied to predict the position of obscure ones and guide to realize local threshold optimizing search. Finally, some edge pixels lying on the edge of reference points are precisely extracted out by gradient extremum search and the reference centers are computed out by ellipse fitting. The proposed method not only ensures the reliable and precise extraction of all the reference points, but also shrinks the processing range and shortens processing time-consuming. Experiments on the SG-II-UP CSF images showed that the method has the features of speed, accuracy and stability, and can be applied in alignment control reliably.