烟幕环境地面红外运动目标快速检测方法

摘要

传统目标检测方法不能很好地满足红外跟瞄系统运动时的地面运动目标快速精确检测需要，另外由于红外烟幕干扰，图像的质量降低，也会影响目标检测的精度。为此，提出了一种烟幕环境下红外运动目标的快速检测方法。该方法首先快速提取两帧红外图像的FAST特征点，提出了改进的二进制BRIEF特征描述符进行背景匹配，然后采用随机一致性算法估计全局运动参数进行运动补偿，最后采用帧间差分法进行目标检测。实验表明，该算法能够有效地降低烟幕干扰对运动补偿的影响，快速精确地检测出运动目标的位置，对烟幕干扰及运动背景具有很好的鲁棒性。%Traditional target detection methods cannot meet the demand of fast and accurate detection while the infra-red tracking and aiming system moves.Meanwhile,infrared smoke debases imaging quality,and reduces the target de-tection precision.A fast method of infrared moving target detection in smoke environment is proposed.Firstly FAST feature points of two frame infrared image are rapidly extracted,and the background is matched by using improved bi-nary BRIEF feature description,the global moving parameters are estimated with RANSAC,and the target is detected with inter-frame difference algorithm.Experiment shows that the algorithm can reduce the effect of smoke on motion compensation,detect moving target quickly and accurately,and has strong robustness for smoke interference and mo-tion background.