Phase recovery from holographic interferometry imagery

Holographic interferometry (HI) is an extremely high-resolution optical metrology technique frequently used for deformation measurement and vibration studies. The physical basis of a HI system is the interference of coherent radiation. The phase difference between two wavefronts, one modulated by a vibrating object, is captured as an interference or fringe pattern. The displacement field is implicitly contained in the fringe image as the phase differences. Thus, to obtain the displacement field. the phase difference map must be recovered in some manner. Various methods for phase map recovery have been developed. One such method involves indirect recovery from the fringe intensity image. For the second group of methods, special geometric imaging configurations and signal processing algorithms are exploited for direct phase map recovery.; Existing methods for indirect phase recovery are based upon a search of the fringe image for local intensity maxima that delineate 2pi radian contour lines of the underlying phase map. However, such algorithms do not perform well in the presence of noise and non-linear fringe contrast variation. A new algorithm developed for this dissertation exploits the topological relationships between the intensity image fringes and their maxima. Specialised filtering techniques, morphological image processing and computer algorithms are utilised to complete the task.; The methods of direct phase recovery, while providing greater resolution displacement field maps, suffer from a discontinuity problem. The numerical processing involved in the recovery has the effect of wrapping the recovered phase. That is, continuous phase maps are rendered discontinuous on the range [-pi,pi] with 2pi discontinuities. The task of phase unwrapping is to remove these discontinuities and thereby obtain the continuous phase map. Existing methods treat the problem of two-dimensional unwrapping and a series of independent, one-dimensional unwrapping tasks. A new approach presented in this dissertation utilises topological definitions of phase maps, multi-scale edge detection and computer vision techniques to perform the unwrapping. Also, by approaching the problem as a two-dimensional task, the final step of the algorithm, the unwrapping, is in fact collapsed to a one-dimensional problem.