The Research Of Ellipse Detect Algorithm Used To The End Face Of PCF

At present, the research of Photonic crystal fiber has become the hot topic athome and abroad. In addition to the study of new fiber optic equipment, moreresearch has changed into the direction of applied exploration. Its application hasbeen involved in communication, imaging and the biomedical sciences. With theapplied further research of photonic crystal fibers, the welding technologybecomes a bottleneck. Because the interior was arranged by many cycles ofhollow fiber, a lot of tiny air hole of the two roots fiber needs accurate alignmentin welding, and to achieve this objective. It needs the identification of photoniccrystal fiber’s end image. Therefore, this topic research is the identificationalgorithms of photonic crystal fibers section image.Firstly, the image preprocessing methods is described, and makes analysisof the characteristics of the technology in various scenarios text de-noising andbinary, sharpening, etc, then selecting suitable detect algorithm for the fiber endimage. And make researches of edge detect method used in optical fiber air holewelding, according to the characteristics of optical fiber cross-section image,Canny operator is used to get pixel level of image edge information in detectingfiber cross section; in order to make detection accuracy, citing improvementhigher Zernike orthogonal moment is used to get sub-pixel levels of edge forfurther treatment, make full use of its rotation invariant features to makeinformation redundancy to achieve optimal and improve the speed.Secondly, make research of the method to identify the fiber air hole centerposition, which is based on the analysis of traditional Hough transform andimproved algorithm. This paper presents a elliptic detective algorithm based onthe Hough transformation: on the basis of sub-pixel edge by means of improvedleast-square fitting operations, to get parameters within the scope of certainrough elliptic information, and then put forward the elliptic detection into a mature algorithm based on the midpoint string round detection, then Houghtransform is applied to the inscribed and circumscribed circles, get the concentriccircles for Hough accumulation, to get the concrete elliptical parametersinformation.Lastly, the main programs are compiled by C++language and OpenCV, theelliptical detection of optical fiber cross section is realized, the results ofexperiment show that method greatly reduces the dimensions of space andinvalid sampling, realize the ellipse quickly and well.