为了有效地提高目标的捕获概率 ,进行红外图像目标跟踪时 ,必须预测目标的质心和姿态 .采用CMAC估计器可以有效地解决常用预测算法运算量与预测精度之间的矛盾 .CMAC估计器先用线性模型进行训练 ,然后预测目标的质心和姿态 .CMAC每帧用实际误差训练一次 ,以减少其预测误差 .CMAC具有很高的预测精度 ,且运算量少 ,适应目标跟踪的实时处理 .实验表明 :CMAC能精确预测目标的质心和姿态 ,适应模型变化 ,具有鲁棒性 In order to effectively improve the target acquisition probability, we must predict the target’s centroid and attitude when using infrared image tracking.Using the CMAC estimator can effectively solve the contradiction between the computational complexity and prediction accuracy of the commonly used prediction algorithms.CMAC estimator Linear model and then predict the target’s centroid and attitude.CMAC each frame with the actual error training time to reduce the prediction error.CMAC has a high prediction accuracy and less computation to adapt to the target tracking of real-time processing.Experiments show : The CMAC accurately predicts the centroid and attitude of a target, adapts to model changes, and is robust