基于非参数背景抑制的弱小目标管线法检测

Dim target detction pipeline algorithm based on nonparametric background removal

介绍了一种红外图像背景抑制的非参数方法(E_kernel)。提出了一种弱小目标的管线检测算法。E_kernel方法不同于传统的线性或非线性背景预测,它对背景杂波分布的统计特性不敏感,受其影响较小,具有非参数特性。管线检测算法对序列图像做若干相同的顺序处理,采用并行分布式计算,处理时间短。仿真试验表明,该算法能有效地检测出低信噪比红外序列图像中的弱小目标的运动轨迹,具有较高的实时性。

A nonparametric background removal method（E-kernel） ot intrared image was discussed. A novel pipeline detection algorithm for dim target was presented. Different with other linear or nonlinear background removal method, E-kernel was insensitive to statistical distribution of background clutter, and had nonparametric property. One of the most important characteristics of pipeline algorithm was its global parallelism. All operations of the algorithm were parallel-distributed, but globally synchronized. Based on simulations of real IR image sequence, this algorithm can effectively detect dim target, and obtain its trajectory. It is adaptable to real-time image processing and target detection.