High resolution wave-front reconstruction based on Fourier transform

VMM(vector-matrix-multiply) method was widely used for wave-front reconstruction in Shack-Hartmann wave-front sensor. Taking advantage of great number of calculations, VMM can give accurate enough result for most of cases, but it can not work properly in scenes which have strict requirement of frequency in time or special domain. In this paper, wave-front is intended to be reconstructed by FTR(Fourier transform reconstruction) method. A common software routine was designed and implemented based on Shack-Hartmann sensor model and FTR algorithms. With the help of these codes, the time analysis of FTR was obtained by software tests. Two kinds of data set were used for the study of accuracy performance, one was ideal slopes derived for known wave-front, the other was slopes corrupted by random noise from actual experiment. Then wave-fronts reconstructed by VMM and FTR were compared and analyzed. Results demonstrate that FTR algorithm has the capability to yield unbiased reconstructions, and shares better time performance than VMM method. Due to extra steps to maintain periodical property of slope data, the wave-front difference between FTR and VMM appears more apparently in edge area. The study and analysis make the limited factors clear, and will guide the implement of FTR in Shack-Hartmann wave-front sensor and supply clues for the improvement in future works.