Enhanced Robust Phase Correlation Based Sub-pixel Feature Matchinig for Target Motion Estimation

This paper presents an enhanced robust phase correlation (ERPC) algorithm for sub-pixel feature matching and its application in target motion estimation for aerial video surveillance. The ERPC can cope with very large motion measurement on the one hand and improve the sub-pixel accuracy by entirely avoiding the ill-posed problem of 2D phase unwrapping in 2D fitting technique of phase correlation on the other. The key advantage of EPRC is its robustness and sub-pixel accuracy which are essential for precise target speed measurement. Furthermore, EPRC is solely applied to the certain parts (not to the whole scene) of a scene where the moving targets are detected, which greatly improves robustness and computing speed of the EPRC based target motion estimation. Finally, we introduce our robust camera compensation and moving target detection scheme. With this simple scheme, we are able to efficiently estimate the motions of multiple targets at sub-pixel accuracy.