روش پردازش تصویر برای پارامترهای هندسی فیبر چند هسته ای

• We propose a novel method for multicore fiber geometric parameters.• The processing include logical noise control, image segmentation, real-time geometric fitting.• We separate the core and cladding by shear function, and detect geometric parameters respectively.• Five different edge detection operators have been compared, and obtained a optimized fitting method.An image processing method has been developed to obtain multicore fiber geometric parameters. According to the characteristics of multicore fiber, we using MATLAB to processing the sectional view of the multicore fiber (MCF), and the algorithm mainly concludes the following steps: filter out image noise, edge detection, use an appropriate threshold for boundary extraction and an improved curve-fitting algorithm for reconstruction the cross section, then we get the relative geometric parameters of the MCF in pixels. We also compares different edge detection operator and analyzes each detection results, which can provide a meaningful reference for edge detection.