由于复杂、快速变化背景的干扰，空中平台红外探测系统下视背景弱小运动目标检测存在巨大挑战，尚无高效、可靠的检测方法，这已成为当前制约空中平台红外探测系统实际发挥效能的技术瓶颈之一。本项目综合利用序列图像中时域及空域信息，研究基于时-空域关联的下视背景红外弱小运动目标高效检测方法：在探究帧间几何变换模型的基础上，采用帧间配准与差分的方法抑制帧间重叠区域杂乱背景，对于帧间非重叠区域杂乱背景，基于空域相关性实现高效抑制；综合利用目标运动速度及灰度外观特性设计轨迹度量准则，并引入批处理模式及迭代更新策略，设计并发式轨迹搜索方法，在背景残差、虚假目标等大量干扰存在条件下，高效地检测真实弱小目标。本项目的开展对于提升空中平台红外探测系统对下视背景弱小运动目标的探测能力具有重要的理论意义与工程应用价值。

In downward looking application scenarios of aerial infrared detection system, there are great challenges in detecting dim moving target under the complex and fast changing background. And there is no efficient and reliable method. This has become one of the bottleneck technologies that restrict the actual performance of the aerial infrared detection system. Using image sequence information, this project studies infrared dim moving target detection in downward looking background based on temporal and spatial cues. On the basis of exploring the geometric transformation model between frames, the clutter background in the overlapping area between frames is suppressed by frame registration and difference. And for the clutter background in the non-overlapping area, the spatial correlation is used to achieve efficient suppression. The speed and the intensity distribution of the target are used to measure the target trajectory reliably. Batch processing and iterative update are adopted to achieve efficient trajectory searching. The development of this project has important theoretical significance and engineering application value for improving the detection ability of aerial infrared detection system for dim moving target in downward looking background.