Motion estimation in the presence of illumination variations

摘要

We consider motion estimation between images of a video sequence in the presence of illumination variations in the scene. The standard model of motion between consecutive images is extended to a prediction model with multiplicative prediction coefficient. This additional coefficient is interpreted as an illumination variation parameter. A pel-recursive motion estimator is adapted to this new motion model in order to estimate both motion and illumination variation fields. We present experiments on real images containing localised illumination variations and show that the proposed approach allows the prediction error to be largely reduced in comparison with the standard pel-recursive motion estimation algorithm.