基于K-means聚类的光纤环边缘提取算法

K-means clustering-based edge detection and recognition algorithms for f iber rings

光纤环是光纤陀螺中引起的两相向传播光波之间的Sagnac相位差的部件,光纤环如何缠绕将对光纤陀螺的测量精度产生直接影响。本文提出一种基于K-means聚类的光纤环边缘检测识别算法,通过与常见边缘提取方法的比较,发现用K-means聚类算法能够准确的找出初始簇中心,可以比较准确完整的将光纤环边缘轮廓从背景中分割出来,并具有较强的抗噪性。

Fiber ring is a component of FOG causing two counter-propagating waves to produce Sagnac phase difference. How to twist the fi ber rings has a direct impact on the measuring accuracy of f iber optic gyro. A K-means clustering-based fi ber ring edge detection and recognition algorithm is presented in this paper. Compared with the common edge extraction algorithms, the K-means clustering algorithm can accurately fi nd the initial cluster center and excise the f iber ring edge out of the background. The algorithm also has a strong resistance to the noise.