野外复杂背景下红外图像的目标检测

野外复杂背景下红外图像序列目标检测是红外野外监视中的重点问题。大量的背景物增加了目标检测的难度。文中针对野外复杂背景下红外图像序列的特点，提出了一种实用的运动目标检测算法。该算法包括两个处理步骤：首先，在场景配准后利用帧间差图像提取目标的运动信息，并据此进行目标的粗检测；其次，合目标运动在时间和空间上的相关性进行精检测。粗检测的低漏判度和精检测的低误差率保证了算法的可靠性。在检测的同时算法确定了目

IR target detection under open country scene with a complex background

IR target detection is a key problem in monitoring for open country scene with a complex background. The quantity and the target-similarity of the background objects make the detection even harder. A practical IR target detection algorithm is presented here. It includes two processing steps. First, to match the scene, extract the movement information, and then perform the coarse detection based on it. Second, to perform the fine detection based on the time-space relativity of the movement. The low missing rate of the coarse detection and the low false rate of the fine detection guarantee the stability of the algorithm. Both the target and its region location are determined by using the algorithm. The experiment results are shown in the paper.