一个实时的点目标检测算法的设计与实现

Design and Implementation of a Real Time Algorithm for Detecting Point Target in Image Sequences

低信噪比条件下图像序列中运动点目标的检测涉及信息量大、目标运动轨迹组合复杂，是多维图像处理与目标识别领域的一个难点，实时处理更加困难。本文在基于分段复合的序贯检测算法基础上，设计出了使用投影累加与最大值投影有效结合进行分段复合的方法，整个算法层次分明、各层操作简单，适于流水处理。最后，本文还给出了该算法在一个简单的流水结构信息处理机上的实时处理途径，满足了一定的应用需求。

Concerning huge amount of information and complicated combinations of trajectories,the automatic detection of moving point target in the low SNR image sequences belongs to the difficult in the area of multidimensional image processing,even to the real time requirement.Under the framework of sequential hypothesis testing based on stage integration,the paper puts forward a method of stage integration,which efficiently combines summation projection with maximum projection.Only using some simple operations,the entire algorithm is hierarchical and fit for the pipeline process.Specially a real time approach is implemented on a simple pipeline IPM (information processing machine). [WT5”HZ〗