Research On The Filtering And Skeletonization Algorithms Of Optical Interferometry Images

During the rapid development of laser technique and computer technique, Modern optical measurement techniques have been becoming widely application on scientific researches and projects, such as Micro-displacement measurement, wavefront retrive and surface tomography reconstruction, etc. Compared with conventional optical measurement techniques, modern optical measurement techniques offer a number of advantages, such as high sensitivity, high precision and convenient data processing. So they have attracted great experiment and theory research, it emerges in experiments that various measurement methods have been produced, such as laser holographic interferometry method, electronic speckle pattern interferometry method and moire interferometry method, etc. Although various details have been used in these measurement methods, it is necessary to research the data processing models and algorithms of optical interferemety image because physical properties of the measurd are charactered by means of optical interferometry fringe patterns in modern optical measurements. This work mainly focuses on the data processing model and algorithm theory research of optical measurement fringe patterns. Based on the image models, the author has presented several filtering algorithms and skeletonization algoritms for the optical fringe patterns, and the performance of algorithms have been explained based on the detailed fringe pattern analysis. Main achievements are summarized as follows.An interferogram filtering algorithm was presented based on discrete topological analysis. The author found that long computing time and complex post processing techniques were required for the model based on the topological gradient. Based on this, the author presented an improved algorithm in which the topological expansion theory was considered. The author converted the disturbing idea according to the pixel in the traditional topological gradient model to the distrubing idea accoring to the edge of the pixel, and applied it to the interferogram filtering problem. Numerical experiments and real interferogram analysis indicate that the filtering model and algorithm can be used in fast interferogram filtering processing and the interferogram with low noise filtering processing.An wrapped phase pattern filtering algorithm was presented based on Bayesian evaluation and Complex-valued Markov Random Field (CMRF). The author found that obscure phenomenon was likely to occur for the wrapped phase patterns with dense fringe distribution in traditional isotropic model based on Bayesian evaluation and CMRF. Considering this, the author presented an improved filtering model and algorithm based on Bayesian evaluation and CMRF, in which the denoising image can can be obtained by means of local elicitation information and combined with the characterization of the wrapped phase patterns, and analyzed the performance of the algorithm by numerical experiments and real interferogram processing. The algorithm can be used to the wrapped phase patterns with high noise and dense fringe distribution, as well as applied to the wrapped phase patterns with sparse fringe distribution.Two skeletonization algrithms of the optical interferogram were presented based on the gray-level data processing technique. Firstly, the author studied the feature of the traditional peaking tracking method (PTM) and proposed an improved PTM based on the goal that detected initial skeleton points of the interferogram with circle fringe distribution, but, it can be used to detect initial skeleton points of the interferogram with complex fringe distribution. Secondly, the author proposed another skeletonization algorithm base on oriented Gradient Vector Flow Fields (GVF), in which the fringe orientation angle can be acted as a powerful aid to perfect the process. Experiments indicate that the algorithm can be used to detect skeletons of the interferogram with dense fringe distribution.The author studied application of skeletonization algorithms for detecting skeletons of wrapped phase patterns, and an experimental apparatus for data acquisition of interferogram was desighed and established. Above-proposed two skeletonization methods were applied to the real interferogram processing, detailed experiments showed the feasibility and performance comparison of the algorithms, and a real-time tracking method of the interferogram with circle fringe distribution was proposed by combining oriented GVF method and circle fitting method.