一种利用符号算法的自适应图象位移估计算法

An Adaptive Image Displacement Estimation Algorithm Using Sign Algorithm

本文提出了一种利用符号算法(Sign Algorithm)的自适应图象位移估计算法,这种迭代算法能直接估计出运动图象序列中两祯目标图象的位移。本文还对算法的性能进行了详细的理论分析,得出估计方差与信噪比、图象相关函数等参数之间的定量关系式,以及由于坐标量化而引起的算法对图象特征的限制条件。理论分析和模拟表明,它具有较强的抗干扰能力和计算量少,便于实时实现等优点,可应用于成象跟踪和图象编码等领域。

A new algorithm for image displacement estimation has been described in this paper. The algorithm, utilizing sign algorithm, can obtain the displacement estimation between two frames of image directly. Having been theoretically analysed, a relationship among the variance of displacement estimation and some parameters such as SNR, correlation of image and window size is developed and a limit-cycle-free condition is given. The theory analyses and simulation results indicate that this estimation algorithm has advantages of strong capability of anti-interference and small burden of computation. Consequently, it may be applied in image tracking and coding.