CAJ | 학술논문

　　复杂地面场景下的红外目标易受背景影响并经常出现遮挡情况，难以简单地依靠亮度或梯度信息检测并跟踪目标。根据复杂背景下红外运动目标与背景的速度场差异，提出了利用光流对目标进行跟踪的算法。首先对图像进行配准，保证在随动跟踪时背景的相对静止；然后在目标的跟踪波门内计算改进的Horn-Schunck光流；最后根据目标的光流特征，优化粒子滤波算法中粒子的转移概率，实现对目标的稳健跟踪。实验结果表明，该跟踪算法能对复杂地面场景下的红外运动目标持续跟踪，并不受目标被短时遮挡的影响。
　　복잡지면장경하적홍외목표역수배경영향병경상출현차당정황，난이간단지의고량도혹제도신식검측병근종목표。근거복잡배경하홍외운동목표여배경적속도장차이，제출료이용광류대목표진행근종적산법。수선대도상진행배준，보증재수동근종시배경적상대정지；연후재목표적근종파문내계산개진적Horn-Schunck광류；최후근거목표적광류특정，우화입자려파산법중입자적전이개솔，실현대목표적은건근종。실험결과표명，해근종산법능대복잡지면장경하적홍외운동목표지속근종，병불수목표피단시차당적영향。
The infrared target under complex ground scene is vulnerable to the background and is often occluded. It is difficult to detect and track targets simply rely on brightness or gradient information. According to the difference of velocity field between the moving target and the complex ground, a new target tracking algorithm with optical flow was proposed. Firstly input images were aligned to ensure a relatively static background in the servo track; secondly the improved Horn-Schunck optical flow was calculated in the tracking gate; lastly the transition probability of the particle in the particle filter algorithm was optimized with optical flow features of the target. Experimental results show that the proposed tracking algorithm can keep track of the infrared moving target under complex ground scene, and is not subject to the impact of short-time occlusion.