约束抽样Hough变换的光纤端面精确定位

为提高经光纤传输后光束方向的稳定性,提出了一种新的光纤端面精确定位方法。该方法针对光纤端面光能量中强模间噪声干扰的特点建立了约束抽样Hough变换,可实现对噪声的去除和对轮廓的准确提取,并根据轮廓的特征把特征点分为基元类和噪声类,对两类点分别进行约束抽样Hough变换,可显著减少存储所需要的内存空间,有效抑制噪声对轮廓的干扰,准确得到属于轮廓的特征点。根据各特征点对轮廓的贡献,采用弦长加权拟合法对轮廓点进行处理,从而得到轮廓的中心位置。实验结果表明,该方法对光纤端面的定位不确定度优于0.2 pixel,成像物镜放大倍数为10时,定位不确定度优于0.4 μm。

Precise location for optical fiber surface with sampling restraint Hough transformation

In order to improve the direction stability of light beam from optical fiber, a novel location method for optical fiber surface was proposed. The energy distribution at optical fiber surface was analyzed and Hough transformation with sampling restraint was proposed in this method to eliminate noise and obtain the contour points according to the contour character. The points divided into two classes(primitive class and noise class) were respectively transformed into parameter space to reduce memory and restrain noise to obtain the contour points. The points were set different weight for their different effect on the center of the contour in fitting process. The experimental results show that the uncertainty is less than 0.2 pixel(less than 0.4 μm) when the proposed method is used to locate optical fiber surface and a ten-fold image lens is used.