针对复杂天空背景条件下低信噪比的红外弱小目标检测问题,提出了一种基于对称差分和光流估计相结合的目标检测算法。对序列红外图像做对称差分运算,通过图像差减运算和自适应阈值分割提取目标可能的运动区域,并对区域做扩张和叠加处理,得到连续帧间目标可能出现的区域。计算每个区域红外图像的光流场,对光流场进行阈值分割,辅以数学形态学滤波等方法,检测区域中的目标。该算法充分利用对称差分运算计算量小和光流检测准确度高的特点,在保证检测准确度的同时大大减少了目标检测算法的计算量。实验及结果分析表明,基于对称差分和光流估计的目标检测算法能实时有效地检测出复杂天空背景下的红外弱小目标。

In connection with the infrared target detecting under complex backgrounds,a detecting algorithm based on symmetrical displaced frame difference (DFD) and optical flow estimation is put forward. Firstly,through using symmetrical DFD algorithm on infrared sequence frame,possible target areas are extracted by using frame difference (FD) algorithm and auto-adaptive segmentation of threshold. The area is extended and piled up,then the possible target area is gotten in continuous frames. Secondly,every infrared frame′s optical flow field is calculated,and through threshold segmenting and combined mathematical morphology algorithm targets are detected in areas. This method can sufficiently utilize the advantages of little calculation of symmetrical DFD algorithm and high exactness of optical flow detecting method. So it can cut down the calculation as well as satisfy the veracity. The experimental result shows that this method works perfectly and can effectively detect infrared targets under complex backgrounds.