论文部分内容阅读
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。
Aiming at the problem of real-time moving targets detection in infrared early warning and tracking system, a new detection algorithm based on high-order statistical criteria is proposed based on analyzing the non-stationary Gaussian characteristics of infrared grayscale images. Firstly, a space-based albino-to-mean filter is used to suppress the spatial background, and an uncorrelated Gaussian background is established for the next high-order statistical criteria in the time domain. Based on the higher-order cumulants above the third order, the “blind” Principle, using high-order cumulants as binary statistical criteria to detect the weak moving targets in the background of infrared images. The algorithm comprehensively considers the characteristics of infrared grayscale images and targets in the time domain and airspace, and greatly enhances the target signal-to-noise ratio. The actual infrared target detection of the earth background shows that this algorithm can effectively detect small moving targets with low signal-to-noise ratio from complex background with less false alarm rate and strong anti-noise ability. The algorithm is easy to implement in hardware and can be effectively applied in the real-time target detection of infrared search and tracking system.