A motion estimation algorithm based on interframe differences detection model for virtual edges detection

摘要

The traditional block matching algorithm in motion estimation costs much time because of the high computation complexity. which causes the deterioration of the real time performance. A new motion estimation algorithm based on interframe differences detection model is proposed in this paper. The new concept of virtual edges is advanced and introduced into the event of the small object moving. The interframe differences detection model for virtual edges detection is then made up. The parameter in the practical model selection strategy is also determined. The experimental results and the analysis of the algorithm complexity show that the new algorithm can not only work well for small moving object direction detection under the complex background, but also has the better real time performance compared with the traditional block matching algorithm.