Integrated approach for estimation and restoration of photon-limited images based on steerable pyramids

In disciplines like fluorescence microscopy and astronomical imaging, the imaging process is based on detection of photons. Fluctuations in photon counting processes are described by Poisson statistics. In this paper, a new combined method based on steerable pyramids is proposed for the estimation of the degradation parameters (like noise and blur) and the restoration of photon-limited images. It consists of the following steps: in the steerable pyramid domain, a noise suppression step is performed, followed by a blur estimation. As a last step, the Richardson-Lucy deconvolution algorithm is applied. These steps are iterated. The only free parameter in the algorithm is the number of iterations, but an empirical stopping rule is suggested in terms of the blur estimation. Otherwise, this method is fully automatic and provides very nice restoration results.