A high-speed accurate system for phase denoising and unwrapping

• Main Author: Gao, Wenjing
• Other Authors: Qian Kemao
• Format: Theses and Dissertations
• Language: English
• Published: 2012
• Subjects: DRNTU::Engineering::Computer science and engineering::Computer systems organization::Special-purpose and application-based systems; DRNTU::Engineering::Computer science and engineering::Mathematics of computing::Numerical analysis; DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision
• Online Access: https://hdl.handle.net/10356/50787
• Institution: Nanyang Technological University

Two-dimensional (2-D) fringe pattern analysis algorithms have been developed for several decades. They impel the frontier interferometric techniques such as synthetic aperture radar, interferometric synthetic aperture sonar, magnetic resonance imaging, diffraction tomography and optical measurement being widely used in more and more measurement applications. Due to the factors such as noises caused by acquisition mechanism and speckle effect, subsampling, and long discontinuities, the modern 2-D fringe analysis algorithms for phase denoising and unwrapping are complicated and usually time consuming. The windowed Fourier transform (WFT) based algorithms including the windowed Fourier filtering (WFF) and the windowed Fourier ridges (WFR) are among the effective fringe analysis algorithms. Along with quality-guided phase unwrapping algorithm, the WFT based algorithms can successfully reconstruct the phase by removing the noises and effectively identify the long discontinuities in the fringe patterns. Despite the merits of the WFR algorithm, a small phase bias is produced by the WFR algorithm for phase estimate. In this dissertation, phase compensation methods are proposed for the WFR algorithm, which is denoted as the WFRC algorithm. Then the theoretical MSEs for local frequency and phase estimates by the WFRC algorithm are derived in the first order perturbation perspective and compared with the corresponding Cramér-Rao bounds to show the efficiency of the proposed algorithm. Another drawback of the WFT-based algorithms is their long computation time, prohibiting them from real-time applications. However, the WFT-based algorithms are highly parallelizable, which implies the potential of being efficiently accelerated by the modern parallel computing hardware. Thus in this thesis we explore using parallel hardware to accelerate the WFT-based algorithms and propose a high-speed heterogeneous system based on multicore CPU and graphics processing units for phase denoising and unwrapping using the two algorithms. The system can be easily integrated into a fringe analysis system.