In this paper, the spatio-temporal gradient method based on the integral projection is proposed and the other tracker-related algorithms to cope with the real tracking situation are investigated. The proposed method can reduce greatly the computational load of the conventional motion estimation algorithm. It offers sub-pel accuracy without complex interpolation process under noisy condition, especially the additive noise image. With one-dimensional spatio-temporal gradient method based on the steepest descent algorithm, measuring range is expanded and convergence rate is faster than the conventional STG method. To improve the overall tracking performance, an optimal window sizing with the intensity gradient function and an integral projection technique with multiple projection axes are introduced.
In spite of the great reduction of computational load, experimental results show that the proposed method has comparable performance to the conventional one under severe tracking condition.