Please use this identifier to cite or link to this item:
|關鍵字:||ESPI;電子斑點干涉術;Enhancing the Contrast of Correlation fringe;Minimum LP-Norm Phase Unwrapping;條紋對比之提昇;相位展開技術;區塊接合||出版社:||機械工程學系||摘要:||
相位展開技術在後處理過程中更是重要的一環。尤其當處理對象為具有高雜訊，如電子斑點干涉術所得之結果，則相位展開技術必須要更為強健才能展開正確。經過本實驗室多年來的研究，對於高雜訊所造成的不連續現象(inconsistency)已有能力克服之；但對於原始相位本身就存在的不連續現象（例如：剪切平面），基本上比雜訊所造成的不連續現象更加不容易執行相位展開之程序。本論文以Dennis C. Ghiglia和Louis A. Romero在1996年所發表的Minimum LP-Norm Phase Unwrapping為基礎，針對原始相位本身就存在的不連續現象加以克服，並對此相位展開法之缺點，提出『區塊接合』理論，進而改善運算效率且提昇對雜訊之免疫能力。
Electronic speckle pattern interferometry (ESPI) has been used to elucidate the in-plane and out-of-plane displacements of an object. It is a method for measuring the deformation or displacement of the surface of an object by recording at least two speckle patterns, one before and one after the object is deformed. By adding, subtracting or multiplying the speckle patterns, correlation fringe patterns with poor signal to noise ratios are obtained. In general, the contrast of the correlation fringe patterns is enhanced using digital filter methods. However, digital filter methods cannot remove the speckle noise efficiently and sometimes leads to inaccurate digital filtering when the noise is intense. This thesis is based on Zi-Neng He's thesis (in 2001) and further cooperated with several fringe and image analysis methods to enable the achievement of correlation fringes with excellent contrast. He's thesis is based on a trigonometric operation to adjust and unify the intensities of all pixels of an interferogram (and is termed as normalization by him.) His method differed from most digital filter methods. An intercomparison of them shows that normalization method outperforms digital filtering method.
In addition, this thesis studies the phase unwrapping technology for ESPI maps with physical shear. In 1996, Dennis C. Ghiglia and Louis A. Romero published the minimum LP-norm 2-D phase unwrapping algorithm to restore the phase surface with real shear. It is unfortunate that the solution could fall if the solution has many local minima. Some of the local minima could be close to an actual global minimum and could thereby yield a solution that is useful for the application at hand. This unfortunate circumstance stems, in part, from the fact that there is no unique global minimum for L0-norm problem. This differs from the uniquely solvable L2-norm problem. Convergence to a local minimum is guaranteed; convergence to a global minimum is not. This disadvantage makes it difficult to analysis the wrapped phase map in ESPI and to obtain the phase distribution correctly. Hence, I propose an innovative method based on the submap-stitching rule and the branch-cut algorithm to circumvent the aforementioned main drawbacks. The computational simulations and experimental implementations of the thesis show the effectiveness of the proposed method.
|Appears in Collections:||機械工程學系所|
Show full item record
TAIR Related Article
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.