一种新的用于点目标检测与跟踪的闭环pipeline算法

A New Closed-Loop Pipeline Method for Point Target Detection and Tracking

提出了一种由滤波管道、检测管道构成的闭环管道算法用于检测与跟踪复杂云天背景下的点目标.引入了基于单帧局部对比度门限的预处理算法进行杂波抑制,设计了基于三帧图像相关性分析的连续滤波器进行噪声剔除.实验分析表明,该算法不仅能够检测出作任意轨迹运动的点目标,而且能够有效地预测跟踪窗.它在搜索空间的缩小、计算复杂性的降低以及杂波剔除方面优于传统的管道算法.

A new method, namely the clos ed-loop pipeline, is proposed for point target detection and tracking in cloudy background. The two pipelines, a continuity filter pipeline and a test pipeline , are employed in the proposed method. At first, a preprocessing step, i.e., a l ocal contrast thresholding, is used to detect targets, then the continuous filte r using three frames correlation analysis is used to effectively remove noise. T he experimental results show that the proposed method can detect and track dim p oint targets with arbitrary trajectories accurately, and can also predict the se arching windows efficiently. The method is shown to be advantageous over the tra ditional one in terms of searching space, computational complexity and clutter r esistance.