基于神经模糊网络的自适应α-β-γ目标跟踪算法

An adaptive α-β-γ tracking algorithm based on neuro-fuzzy network

针对传统α-β-γ滤波器噪声统计特性未知时,受色噪声的影响精度严重降低,甚至出现发散等现象,设计了一种基于神经模糊网络的自适应的α-β-γ滤波跟踪器。该滤波器通过利用神经模糊网络作为误差估计器,估计出α-β-γ滤波器的估计误差,从而对α-β-γ滤波跟踪器的预测结果进行修正,得到更优的预测值。通过计算机仿真以及在图像目标跟踪平台验证结果表明,该算法可以克服传统算法的局限性,有效地防止滤波器发散,缩小实际的滤波误差,提高滤波精度,实现对跟踪结果的在线改进。

Traditional α-β-γ filter has the problems of degradation or even divergence in tracking targets for nonlinear systems with colored noises.An adaptive α-β-γ tracking algorithm based on neuro-fuzzy network is proposed in the paper.The estimation error is obtained online to modify the filtered result by taking neuro-fuzzy network as the estimator.The simulation results and engineering implementation indicates that the proposed tracking algorithm can restrain colored noise and eliminate the divergence.It can significantly reduce the error of traditional algorithm and improve the tracking accuracy of the system.