Measurement of the point-spread function of a noisy imaging system

Christopher D Claxton,Richard C Staunton
DOI: https://doi.org/10.1364/josaa.25.000159
Abstract:The averaged point-spread function (PSF) estimation of an image acquisition system is important for many computer vision applications, including edge detection and depth from defocus. The paper compares several mathematical models of the PSF and presents an improved measurement technique that enables subpixel estimation of 2D functions. New methods for noise suppression and uneven illumination modeling were incorporated. The PSF was computed from an ensemble of edge-spread function measurements. The generalized Gaussian was shown to be an 8 times better fit to the estimated PSF than the Gaussian and a 14 times better fit than the pillbox model.
What problem does this paper attempt to address?