Multidirectional wavelet-based fringe analysis for surface diagnosis by optical profilometry

Abstract Scientists, engineers, and manufacturers have a critical need for better techniques of diagnosis and quality control. The optical metrology uses the practical light and computer sciences, in order to perform simulation, design, computing and inspection for many scientific and industrial applications, such as optics, mechanics, aeronautics, electronics and many others. The fringe pattern analysis is a method to perform some operations on the optical images, in order to get an interferometric phase map and then to extract some useful information from it. In this paper, an exciting improvements of local fringe demodulation algorithm are presented, which are based on a new multidirectional wavelet. The numerical and experimental works show interesting gains compared to the other standard algorithms. Our approach executes as fast as the popular phase retrieval methods, but with substantially improved accuracy in demodulating noisy fringes. All of this comes without any pre-processing by filtering models.