基于时空分集的弱点状动目标高性能实时检测技术研究

A spatial-temporal diversity based real-time high performance TBD technique for Detection of Dim Moving Point Targets in Image Sequences

降维处理法在多维信号检测中虽然减少了计算量,简化了检测过程,但是将会导致系统检测性能的下降,这一点是实际应用所不允许的。在序列图像情况下,为提高算法的实时性,研究了一种基于时空分集理论的降维处理及实时检测技术:时间分集实现了从时空三维图像向二维空间的投影,从而目标运动轨迹的搜索空间维数从时空三维变成了二维;空间分集完成了目标能量的积累,从而提高了信噪比。为提高系统检测性能,对决策统计量的分布函数进行了变换,改造了分布形状,获得了有利于提高检测概率的分布。给出了具体实施方案,相应的理论分析和实验结果。

A probability distribution function transformation based technique for the improvement of system detection performance and a spatial-temporal diversity based technique for dimension reduction and real-time detection of dim moving point targets in IR image sequence is presented. The theoretical analysis and simulation results are also given in this paper.