An implementation of target motion estimation based on target edge segment features for vehicle monitoring applications

In this paper we present an implementation of a novel motion estimation algorithm for vehicle targets. The algorithm is based on the analysis of inter-frame correlations of the target edge segments, and is capable of estimating both translational and rotational motions. Through the analysis of the edge segment feature correlations, the distributions of the inter-frame motion probability are calculated and accumulated into the global motion spaces, from which the inter-frame target motion can be estimated. This paper introduces the theory, the implementation, the full experimental results and the advantages of this algorithm. Experiments show that it may be a potential approach to the motion estimation of vehicle for applications in some situations.