Linear filters for deconvolution microscopy

We present two linear, non-iterative approaches for deconvolution of three-dimensional images that are able to produce good approximations of the true fluorescence concentration in computational optical sectioning microscopy. Both the proposed filters take into account the nature of the noise due to the low level of photon counts. We present some results of the applicability of the methods using a phantom image, where the improvement in signal-to-noise ratio was used in order to quantify the restoration results, and also using real cell images. We compare the algorithms with the regularized linear least squares algorithm considering different levels of Poisson noise.