基于红外成像的小目标检测技术研究

摘要

针对远距红外弱小目标的成像特点，提出了一种基于连续帧的红外成像小目标先检测后跟踪算法。首先利用增强型简化高通滤波预处理提高图像信噪比，然后根据直方图迭代得到的自适应阈值进行二值化处理和目标标记，在后继的连续帧中根据导弹工作状态不同而确定不同的选择截获算法，判断截获后转入小目标对比度跟踪阶段。仿真实验结果表明，本文算法在复杂背景、低信噪比条件下，仍能准确检测出远距小目标，阈值选取无需人工干预，检测速度快，利于弹上实时应用。%According to the infrared imaging trait of the small target in long distance , a small target detecting and tracking algorithm with continuous frames is presented .Firstly, it uses the enhanced simpli-fied high-pass filter to elevate SNR .And after binarization is used with the threshold got from iteration of histogram, the targets are marked and their characters are extracted .Finally, the aimed target is con-firmed after several consecutive frames and the process turns into contrast tracking with the target ’ s cen-troid as aiming point .Tests show that the algorithm presented can realize fast detecting and stable tracking the small target under complex backgrounds , and the binarization threshold is got without manual work . The rate of this algorithm is fast , so it is convenient for using in the air-to-air missiles.