“…Moving object selection algorithms, on the other hand, first detect motion regions by inspecting the inconsistencies in the input pixel intensities and then seek to eliminate the corresponding artifacts by using samples from either a single input LDR image [JLW08,LC09,RWPD10] or a subset of the input images that are found to be consistent [GGC * 09, RKC09, RC11, HLL * 11, SSM12, OLK13, SPLC13, GKTT13]. While some studies make use of pixel-wise optical flow correspondences [KUWS03,ST04,JO12,HDW14], others utilize patch-based dense matching methods [POK * 11, SKY * 12, HGP12,HGPS13]. While some studies make use of pixel-wise optical flow correspondences [KUWS03,ST04,JO12,HDW14], others utilize patch-based dense matching methods [POK * 11, SKY * 12, HGP12,HGPS13].…”