基于SVD背景抑制和粒子滤波的弱小目标检测*

针对云天背景下红外弱小目标的检测算法中常见的目标漏检和检测错误问题，提出了一种基于奇异值分解背景抑制和粒子滤波联合检测算法。该算法首先采用奇异值分解滤波抑制红外图像背景，获取候选目标位置，然后采用粒子滤波算法估计目标运动状态，获取目标搜索窗口，最后将单帧检测候选目标与预测的搜索窗口相结合实现小目标检测。对真实红外图像序列进行实验表明，该方法有效地解决了SVD滤波单帧漏检和粒子滤波预测错误导致的目标检测错误问题，从而提高了低信噪比下弱小目标的检测能力。

Detecting small targets based on SVD for background suppression and particle filter

In order to improve the detection ability for infrared small targets under cloudy sky background, the authors proposed a novel detection algorithm based on singular value decomposition (SVD) filter and particle filter. In this algorithm, firstly, a SVD filter was utilized to suppress the background and obtain several candidate targets. Secondly, particle filter was utilized to track the target and get the target search window. Finally, the target was detected by combining the candidate targets with the search window. Several experiments show that the proposed algorithm can detect the small target effectively when the background is complex or noisy.