基于图像处理的光子晶体光纤截面结构重建及色散研究

发布时间：2018-03-23 23:31

本文选题：光子晶体光纤　切入点：图像处理　出处：《天津工业大学》2017年硕士论文

【摘要】：随着光子晶体光纤的发展和研究,拥有特殊结构和灵活性特性的光子晶体光纤应用越来越广泛,精确测量光子晶体光纤的性能成为了研究重点。针对实际光子晶体光纤由于结构不规则难以进行数值分析的缺点,本文基于图像处理算法对光子晶体光纤扫描电镜截面图像进行处理,并在此基础上提取出光纤的几何结构作为测量其特性的模型,然后在COMSOL Mutilphysics软件上利用重建的几何结构模型结合光子晶体光纤的有限元数值分析方法对其色散数值模拟,提高了光子晶体光纤光学特性的评价准确真实性。本文的主要工作是利用图像处理算法对光子晶体光纤进行几何结构重建,实现精确建模的关键是对光纤截面图像的识别和精确提取。首先对PCF截面图像采用中值滤波和阈值分割等图像处理方法进行伪信息预处理,选用经典Canny算子检测寻找其边缘,然后利用改进亚像素算法筛选出精确的边缘信息,精确定位PCF截面图像的结构边缘;其次利用基于RANSAC算法改进的最小二乘法对图像的特征点进行拟合操作,即用随机抽样法对已提取的图像特征点筛选,剔除几何结构模型中误差较大的特征点后,再使用最小二乘法对筛选后的特征点进行拟合操作,对提取的几何特征拟合出最优几何—椭圆;最后在基于霍夫曼变换的基础上,引入了同心圆椭圆检测算法,利用上述算法能够对截面结构进行很好的识别及提取,建立起最优光纤模型。利用已经得到了 PCF截面的重构模型,在COMSOLMutilphysics软件平台上结合光子晶体光纤的有限元数值分析法进行色散数值模拟,并对计算结果进行误差分析。用这种方法,本文重建了一种实际光子晶体光纤产品的扫描电镜截面图像的结构和分析了其色散性质,数据计算结果符合产品参数。
[Abstract]:With the development and research of photonic crystal fiber, photonic crystal fiber with special structure and flexibility is more and more widely used. The accurate measurement of the performance of photonic crystal fiber has become the focus of research. In view of the disadvantage of practical photonic crystal fiber which is difficult to be numerically analyzed because of its irregular structure, In this paper, the scanning electron microscope cross-section image of photonic crystal fiber is processed based on image processing algorithm, and the geometrical structure of optical fiber is extracted as the model to measure its characteristics. Then the dispersion of photonic crystal fiber is simulated by using the reconstructed geometric structure model and the finite element numerical analysis method of photonic crystal fiber on COMSOL Mutilphysics software. The evaluation of optical properties of photonic crystal fiber is improved. The main work of this paper is to reconstruct the geometrical structure of photonic crystal fiber by image processing algorithm. The key to accurate modeling is to recognize and extract the cross-section image of optical fiber. Firstly, the image processing methods such as median filter and threshold segmentation are used to preprocess the pseudo-information, and the classical Canny operator is used to detect and find the edge of the image. Then the improved sub-pixel algorithm is used to filter out the accurate edge information and locate the structural edge of the PCF section image accurately. Secondly, the improved least square method based on the RANSAC algorithm is used to fit the feature points of the image. The method of random sampling is used to screen the feature points of the extracted images. After the feature points with large errors in the geometric structure model are eliminated, the least square method is used to fit the selected feature points. The optimal geometric ellipse is fitted to the extracted geometric features. Finally, based on the Huffman transform, a concentric circle ellipse detection algorithm is introduced, which can be used to identify and extract the cross-section structure. The optimal fiber model is established and the dispersion numerical simulation is carried out by using the reconstructed model of PCF cross section and the finite element numerical analysis of photonic crystal fiber on the platform of COMSOLMutilphysics. By using this method, the structure and dispersion properties of an actual photonic crystal fiber product section image of scanning electron microscope are reconstructed and analyzed. The calculated results are in accordance with the product parameters.
【学位授予单位】：天津工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：O734;TP391.41

【参考文献】