摘要
通过CCD显微成像系统采集Panda型保偏光纤的截面图像,利用基于Hessian矩阵的亚像素精度边缘检测算法提取两应力圆边缘线点,进行椭圆拟合后过两椭圆中心连线得到偏振轴轴线,实现了偏振轴轴线的自动识别及与已知键定位轴相对角度的自动检测。实验结果表明,在均值为0、方差为0.5的高斯噪声下,本文方法仍可保证偏振轴角度的检测偏差未超过±0.3°,并对真实光纤截面图像的处理具较好的鲁棒性。
The CCD micro-imaging system is employed to get the polarization axis from the cross-section image of Panda PM fiber. The edge points of stress rods are detected by the sub-pixel edge detector based on Hessian matrix. By linking the two centers of the ellipses that are fitted with the sub-pixel edge points, the polarization axis is got. Consequently, not only the automatic detection of the polarization axis but also the accurate calculation of the angle between the PM fiber polarization axis and the orientation key axis are realized. Thus the high accurate alignment of these two axises is ensured. Experiment results indicate that,the precision of the proposed method is less than ±0. 3° while the variance of the Gaussian white noise is 0. 5. Moreover,the real cross-seetion image can be processed very well with this method.
出处
《光电子．激光》
EI
CAS
CSCD
北大核心
2010年第4期565-568,共4页
Journal of Optoelectronics·Laser
基金
国家自然科学基金资助项目(60974108)
航天科技支撑基金资助项目
关键词
CCD显微成像
保偏光纤活动连接器
偏振轴
亚像素边缘提取
CCD micro-imaging
PM optical fiber, connector
polarization axis
sub-pixel accuracy edgedetection
