Comparison of deconvolution, SVRL and RDM on images taken from a point-scanning multimode fiber micro-endoscope. The raw images and the SVRL results are obtained directly from ref. 15 . Top: sample with 1-μm beads; all methods resolve the circularly shaped beads in the center (blue inset). Away from the center, however, deconvolution and SVRL fail to completely remove the ellipticity of the bead but RDM does (green inset). Moreover, in the corner only RDM can resolve bead clusters into their component beads (yellow inset). Bottom: live Wistar rat neuron; RDM can clearly resolve structures (blue inset), sharpen the neuron spine (green inset) and reveal a point-like structure (yellow inset) near the edges, whereas the other methods cannot.
