针对光纤端面制备中各种因素的影响会导致光纤端面产生缺陷的问题, 研究了一种基于机器视觉技术的光纤端面检测系统。针对光纤端面显微图像的结构特点, 使用图像处理算法拟合其中的椭圆, 并提出了一种基于同心椭圆约束的椭圆检测方法, 挑选出纤芯和包层轮廓的拟合椭圆。依据获取的拟合椭圆参数将光纤端面划分为三个检测区域, 进行分区域的瑕疵点检测和统计。全部算法使用VS2010和OpenCV开发实现。实验表明, 该检测系统可以对光纤端面进行准确的质量测定。

In view of the possible defects resulted from the effects of various factors in the preparation of optical fiber ends, this paper develops a machine vision technology-based fiber end detection system. In the light of the structural features of the microscopic fiber end images, it fits the ellipse in them using the image processing algorithm and proposes a concentric elliptical constraint-based ellipse detection method and picks out the fitting ellipse of fiber core and cladding contour. On the basis of the parameters of the acquired fitting ellipse, it divides the fiber end into three sectors of for regional flaw detection and statistics. All the algorithms are developed and realized by using OpenCV function call in VS2010. Experiments show that this detection system can accurately measurate the quality of optical fiber ends.