基于运动特征的远距离红外目标检测方法

在红外热图像序列中检测远距离运动目标具有重要的军事和民用价值.由于远距离目标成像后所占的像素较少,同时受成像环境和成像条件的影响可能会存在较强的背景噪声,目标与背景的差异不明显,检测比较困难.本文针对热红外图像序列,提出了一种远距离目标的检测方法.该方法首先利用top-hat算子选择备选目标,然后利用备选目标在不同帧中的相关程度分析目标的运动特征,选择有确定运动特征的目标作为检测结果.该方法不需要进行背景估计,可以有效地避免强背景噪声的影响;通过调整运动模型,可以应用于不同运动特征的目标检测问题.实验结果证明了本文方法的有效性和鲁棒性.

Long Range Moving Target Detection Based on Motion Analysis

Detection of long range targets in thermal infrared image sequences is of interest in many applications such as military field and surveillance system. Incoming targets at long range where the motion is small and signal to noise is poor are difficult to detect in cluttered thermal infrared image sequences. In this paper, long range target detection in cluttered thermal infrared image sequences is presented. At first, possible targets are detected using top-hat detector in thermal infrared video. Next, motion feature is analyzed by measure the correlation of a target in different frames. Targets with reasonable motion characteristics are decided as long range moving target. In the proposed method background estimation is not needed and heavy noise could be avoided. By modify motion model, targets with different motion features could be detected. Experiments show that the proposed method is feasible and robust.