Blind multi-frame deconvolution for the correction of space-variant blur in images

Abstract: This paper demonstrates a practical method that can correct spatial varying blur from a set of images of the same object. The algorithm jointly estimates the object and local point spread functions~(PSF). The method prioritizes sections with small spatial variation in the PSF for deconvolution. This novel approach can handle large translations in the local PSFs, hence the algorithm is able to correct for morph in the images. Robustness to noise is demonstrated in numerical simulations. Numerical experiments are conducted where the performance of the algorithm is compared to a state-of-the-art method found in literature. The algorithm can be used in situation with space-temporal variation of the PSF and can be applied in situations where the signal-to-noise ratio is low.