Demodulation of fringe patterns corrupted by non uniform intensity variations in dynamics studies

Abstract

Optical interferometric techniques have come to forefront in precision metrology applications such as surface profilometry, deformation analysis and defect testing. Reliable phase measurement from a recorded fringe pat- tern is the primary goal in most of these interferometric techniques. A primary constraint for accurate phase measurement is non uniform intensity fluctuations in fringe patterns caused by irregular illumination, uneven reflectivities and pixel defects. This problem is further aggravated in dynamics studies with multiple image capture where the intensity fluctuations can vary with time and thus lead to several fringe patterns getting affected. In this paper, this problem is investigated using the second order optimization framework along with the total variation regularization on a graphic processing unit (GPU). In our study, a series of fringe patterns are simulated with additive white Gaussian noise at signal to noise ratio of 20 dB. To induce fringe corruption due to the non uniform intensity conditions, the object wave amplitude is varied using an ellipse shaped filter which enables brighter conditions or stronger amplitudes inside the elliptical boundary and relatively darker conditions or weaker amplitudes outside of it. For time varying studies, we varied the size of the filter sequentially to gener- ate a series of interferograms with temporally varying non-uniform intensities. Further, orientations of elliptical illuminations are changed to verify the robustness of the algorithm. For processing a series of interferograms each with size of 512 by 512 pixels with dynamic range of 10 (ratio of brightest to darkest amplitude), we obtained root mean square error less than 0.25 radians for every interferogram using the proposed method. The results show that the proposed method is robust for handling non-uniform intensity corruptions in fringe patterns for dynamics based investigations.