Closed fringe pattern processing techniques : development and comparison

• Main Author: Wang, Haixia
• Other Authors: Qian Kemao
• Format: Theses and Dissertations
• Language: English
• Published: 2012
• Subjects: DRNTU::Engineering::Computer science and engineering
• Online Access: http://hdl.handle.net/10356/50639
• Institution: Nanyang Technological University

Optical interferometric techniques offer non-contact, high accuracy and full field measurement, which are very attractive in various research and application fields. Fringe patterns are the recorded results of the optical interferometric measurement techniques. The complexity of fringe patterns processing, i.e. the complexity of retrieving the measured results, causes inconvenience when using the optical interferometric techniques. The aim of the fringe pattern processing is to retrieve the phase from the fringe patterns. For various types of fringe patterns, the closed fringe pattern is most challenging to be processed. Besides the difficulty of phase retrieving itself, noise especially speckle noise is another encountered problem. In this thesis, the denoising and the phase retrieval from closed fringe patterns are the two main targets. Several contributions have been made to fulfil the closed fringe pattern processing regarding the denoising, phase retrieving and orientation/direction estimation. Denoising: First, a coherence enhancing diffusion (CED) is introduced to the fringe pattern denoising for the first time and produces promising results. Adaptive smoothing is developed in the CED with a special designed distance map. Second, the CED has been analyzed and compared with other existing spatial domain denoising techniques both theoretically and quantitatively. The CED is shown to be the most satisfactory among others. Third, the CED representing the spatial domain techniques while the windowed Fourier filtering representing transform domain techniques are compared. It shows advantages in certain cases when the noise is heavy and the fringe densities are not very high. With these three contributions, we attempt to present a thorough and useful investigation for fringe pattern denoising techniques. Phase retrieving: Speed, robustness and accuracy are three main concerns of the phase retrieving techniques. Two techniques are proposed to demodulate closed fringe pattern. A frequency-guided regularized phase tracker is proposed to increase the success rate of the regularized phase tracker technique and maintain its acceptable high speed. A frequency-guided sequential demodulation with Levenberg-Marquard optimization is proposed to increase the computation speed of the frequency-guided sequential demodulation techniques by about 21 times and maintain its high robustness. The results obtained show that these two techniques are robust, fast and reasonably accurate. The accuracy of these two techniques is further improved by promoting their phase assumption from linear to quadratic, which results in another two proposed techniques: the quadratic phase matching and frequency guided regularized phase tracker and the quadratic phase matching and frequency guided sequential demodulation. Thus, gradually we develop two closed fringe pattern demodulation techniques that achieve high speed, robustness and accuracy.