动态图像跟踪优化算法的研究与仿真

Research and simulation of dynamic image tracking and optimization algorithm

传统图像跟踪算法中,跟踪的图像搜索过程需要历遍所有特质,在图像场景较为复杂的情况下,在"无用"匹配点上耗费大量计算时间,跟踪过程误差较大。提出一种适用于复杂场景下动态图像跟踪优化算法,选择在复杂场景下鲁棒性较强的参数,以增加复杂场景下目标描述的信息量和稳定性;引入一种MAD(平均绝对差)匹配算法:在进行动态图像跟踪过程中采用MAD算法和鲁棒性参数相结合,通过动态帧图像和静态帧对比量即MCD(最多临近点),设定跟踪阈值,通过选取后的图像实现动态图像的运动目标跟踪。仿真实验结果表明,提出方法的跟踪精度对比传统方法有明显提高。

The search process of the tracked image in traditional image tracking algorithm needs to traverse all idiosyncrasies,which expend a large amount of computing time on＂useless＂matching points in complex image scene,and result in big error in tracking process. A dynamic image tracking and optimization algorithm applied to complex scene is put forward. The parameter with strong robustness in complex scene was selected to increase the information quantity and stability of the goal description in complex scene. A MAD（mean absolute deviation）matching algorithm is introduced. The combination of MAD algorithm and robustness parameter is adopted in the process of dynamic image tracking. The tracking threshold is set by the contrast quantity（MCD）of the image static and dynamic frames. The selected image can realize the moving target tracking in dynamic image. The simulation experiment results show that the tracing precision of the proposed method is improved more obviously in comparison with that of traditional method.