Geometry optics principal based image sequence registration method

Gray gradient based image registration method is one of the most fundamental and effective methods of optical flow, target tracking and image registration. But this method is not robust under the condition that the target has a large motion when applied to image sequence registration. In this study, we incorporated geometry optics principal based motion information into the process of registration to improve image sequence registration accuracy. First the motion trajectory of feature point was fitted based on geometry optics principal, then the trajectory was refined by using variable-sized block matching method, finally the gray gradient based method was used to compute registration parameters. Experiments on synthetic sequences and practical sequences show that the proposed algorithm is effective under practical conditions such that the target moves quickly or the object imaging size changes largely.