Illumination invariant motion segmentation of simply connected objects
Martin Bichsel · 1994
A new segmentation algorithm exploits local image quantities which are invariant to changing illumination. Local object-background probability estimates are obtained by comparing illumination invariant quantities in an actual image with the corresponding quantities in a reference image. The objects' simply connectedness is included directly into the probability estimates and leads to an iterative optimization procedure that is implemented efficiently. This new approach avoids early thresholding, explicit edge detection, motion analysis, and grouping.