Abstract: A new algorithm combined facet model and prediction method is presented to detect infrared small moving target in image sequence.Based on the moving continuity and direction of the target in image sequence，a second-order auto-regressive prediction model of the target search window is developed first；The authors then research the method of fitting the image intensity surface of the search window by cubic facet model，and exploite the detect operators to search maximum intensity points in the window.The possible small target position is determined by the maximum intensity points according to the intensity feature of the target in infrared image.Experimental results with the infrared image sequence show that the proposed algorithm can successfully detect the small target，the real-time and anti-noise performance of the algorithm are better than traditional algorithm.

Key words: infrared small moving target, target detection, cubic facet model, prediction algorithm, target search window

摘要： 针对红外图像序列中运动小目标，提出一种基于预测的小面拟合目标检测算法。首先根据图像序列中目标运动的连续性及方向性，采用二阶自回归模型预测目标的搜索窗口。然后根据天空背景红外图像中小目标灰度高于背景的特征，利用小面拟合模型对搜索窗口的局部区域作灰度曲面拟合，提取搜索窗口内灰度极值点作为目标，提出了相应的目标检测算子。对红外图像序列的实验表明，该算法可有效检测天空背景下红外运动小目标，算法的实时性和抗干扰能力优于传统的目标检测方法。

关键词: 红外运动小目标, 目标检测, 小面模型, 预测算法, 目标搜索窗口