由于在成像制导过程中需要实时处理大量的信息，为了在尽可能保留有效信息的情况下降低计算量，采用了一种人眼视觉非均匀采样模型——对数极坐标模型，来压缩信息量以提高计算速度；另外，由于对数极坐标变换对目标形状具有旋转和缩放不变性，在跟踪非刚性目标时该模型能表现出很好的稳健性；考虑到复杂环境中目标跟踪包含强噪声和强杂波，而且目标模型是非高斯非线性的，针对这些问题，在人眼非均匀采样模型的基础上，将基于目标强度特征的Mean shift 跟踪方法与粒子滤波相结合，将Mean shift 算法用于采样粒子滤波,有效地降低了抽样的粒子数，并对算法进行了数字仿真；实验结果表明，该方法能够有效抑制匹配点漂移，并且降低目标跟踪的计算量，是一种稳健的目标跟踪方法。

Because plenty of information needs to be dealt with during the process of image-guiding, in order to retain effective information as much as possible and to lower the amount of calculation, a human visual model of non-uniformity sampling, Log-polar coordinate model, is used, to compress the amount of information and to improve the speed of computing. In addition, because of the invariance of the Log-polar coordinate transformation on the target shape with the rotation and scaling, the proposed model is very robust in tracking non-rigid targets. Taking the last stage in the image tracking into account, centroid and corners of the class tracking method will produce corner’s drift, in order to suppress corner’s drift, the Mean Shift tracking method is used, which is based on target strength characteristics. Taking the complex environment of the target tracking with noise and clutter into account, and the target model of non-Gaussian non-linear in nature, the human eyes non-uniformity sampling model is used. In our approach the Mean Shift tracking method and particle filter are combined and the algorithm is simulated. The experimental results show that the method can effectively suppress match-points drifting, reduce the calculation cost of target tracking, and is a robust target tracking method.