Computational Imaging at the Infrared Beamline of Australian Synchrotron Using Lucy-Richardson-Rosen Algorithm

Abstract

The Fourier transform infrared microspectroscopy (FTIRm) system of the Australian synchrotron has a unique optical configuration with a peculiar beam profile consisting of two parallel lines. The beam is tightly focused using a 36× Schwarzschild objective to a point on the sample and the sample is scanned pixel-by-pixel to record an image of a single plane using a single pixel mercury cadmium telluride detector. A computational stitching procedure is used to obtain 2D image of the sample. However, if the imaging condition is not satisfied, then the recorded object information is distorted. Unlike commonly observed blurring, the case with Schwarzschild objective is unique with a donut like intensity distribution with three distinct lobes. Consequently, commonly used deblurring methods are not efficient for image reconstruction. In this study, we have applied a recently developed computational reconstruction method called Lucy-Richardson-Rosen algorithm (LRRA) in the online FTIRm system for the first time. The method involves two steps: one-time training and imaging. In the training step, the point spread function (PSF) library was recorded by temporal summation of intensity patterns obtained by scanning the pinhole in x-y directions across the path of the beam using the single pixel detector. In the next step, the process was repeated for a complicated object along only a single plane. Object images without spatial aberrations were reconstructed by processing of the object intensity distribution with the PSF library using LRRA. 3D imaging has also been demonstrated using two thin samples.

Keywords: Imaging, Holography, computational imaging, microscopy, deconvolution, coded aperture imaging