基于背景移除的时域目标检测

针对时域目标检测算法中跟踪数据量大、实时实现难度高的缺点,提出一种基于背景移除的时域目标检测方法.该方法首先根据不同像素点的时域起伏特性建立一个统一的模型,进而利用最小二乘法估计出该模型的参数,实现静态背景的移除.然后采用最小值滤波估计出目标信号的检测基准,并进一步分析了像素时域特性偏离该基准的分布特性,最终得到一个合适的目标检测量度.将所给出的算法应用于实际运动弱小目标的检测,实验结果表明,此算法对于复杂背景下的运动弱小目标具有很好的检测性能.

Temporal filtering for target detection algorithm based on static background elimination

A new temporal based target detection algorithm was presented by using static background elimination to deal with the drawback of large scale data processing and real-time implementation in temporal filtering. Firstly, a uniform model was developed based on the temporal behavior of different types of pixels. And the parameter of the model was es-timated by introducing least square method. Using the new model, static background pixels could be eliminated. Secondly,a minimum filer was designed to obtain the baseline of the temporal profile. Finally, the deviation of the temporal profile and its baseline was analyzed, which led to a detection criterion. Since the target signal was well preserved during the static background elimination and large sale tracking data was eliminated, the method could have higher speed and com-putational efficiency. Experiments with real image sequences have proved the validity of the new approach.