基于高阶统计判据的红外弱小运动目标检测

针对红外预警与跟踪系统中的实时弱小运动目标检测问题，在分析红外灰度图像的非平稳高斯特性的基础上，提出了一种基于高阶统计判据的检测算法。先用一个空域的白化去均值滤波器进行空间背景抑制，为下一步时域高阶统计判据建立一个不相关的高斯背景，根据三阶以上的高阶累积量对于高斯随机过程“盲”的原理，用高阶累积量作二元统计判据检测红外图像背景中的运动弱小目标。算法全面考虑了红外灰度图像和目标在时域与空域方面的特性,大大增强了目标信噪比。通过实际获取的大地背景目标红外数据检测表明,此算法能有效地从复杂背景中检测低信噪比运动小目标,虚警率少,抗噪声干扰能力强。算法易于硬件实现,能够有效地应用于红外搜索与跟踪系统的实时目标检测中。

Infrared dim moving target detection based on high-order statistic judgement

For the real-time detection of small target in the IRST systems, a detection algorithm based on high-order statistic judgement is presented in this paper. The algorithm uses a spatial demeaning whiten filter to establish an irrelevant gaussian backgroud for next temporal high-order cumulative statistics. Based on the fact that the cumulant of three order is blind to Gaussian processes, the infrared dim targets are detected in terms to high-order cumulant as binary statistic criterion. Simulations on real infrared image sequences indicate that the algorithm is more effective in detecting low SNR moving small targets in complex background and in reducing false alarm rate. The algorithm is easily implemented in hardware and can be applied effectively in the real-time target detection in the IRST systems.