保偏光子晶体光纤图像特征孔中心提取算法

Detection of the fast and slow axes is regarded as a key technology in applications of the polarization-maintaining photonic crystal fiber. Aiming at such problems as large amount of air holes, noise and pseudo-edges, a positioning method which combines coarse positioning by bimodal mean binarization method and fine positioning by radial scanning three-point precision method was proposed. The former method was used to realize rough positioning of the center of gravity of air holes. Then, it uses the big air hole to orientate, selects two air hole areas with symmetric structure, removes the coarse error points, determines the three edge points to accurately locate the center of the circle, and uses the radius threshold to reduce the error. Experimental results show that the proposed method can segment the fiber surface image effectively, and the center error is less than 0.2 pixels.