A Sum Square Error based Successive Elimination Algorithm for Block Motion Estimation

Block Motion Compensation is used for finding and compressing temporal differences between frames and estimating the motion between frames. Optimal temporal compression demands minimising the error between blocks and this is best accomplished by having as large a search area as possible. Motion estimation, on the other hand, demands a small search area to preserve the coherence of the motion vector field. This paper develops a lower bound on the Sum of Squared Errors (SSE) in terms of the energies of the candidate best match and the search blocks and describes a fast method for calculating the energies of overlapping blocks. The search for a best matching block can then be truncated by eliminating search positions which do not satisfy this bound. The proposed algorithm is compared to an exhaustive search. The results show a reduction in search time and identical motion fields.