CAJ | 학술논문

针对云背景下红外小目标检测对背景起伏依赖性强的问题，提出采用分形方法实现不同起伏背景的小目标检测。首先利用分数布朗随机场模型对实际云背景图像建模得到分形曲线，然后通过分析小目标存在对云背景分形特性的影响，总结出云背景区域含小目标时其分形曲线面积增大的规律，据此提出了基于分形曲线面积差量的小目标检测方法。实验结果表明：在虚警率为10－3时，分形方法对弱起伏和中等起伏云背景下红外小目标检测概率达到1.0，对强起伏云背景下红外小目标检测概率达到0.6，较现有其他一些方法具有更好的检测效果和背景适应能力。
침대운배경하홍외소목표검측대배경기복의뢰성강적문제，제출채용분형방법실현불동기복배경적소목표검측。수선이용분수포랑수궤장모형대실제운배경도상건모득도분형곡선，연후통과분석소목표존재대운배경분형특성적영향，총결출운배경구역함소목표시기분형곡선면적증대적규률，거차제출료기우분형곡선면적차량적소목표검측방법。실험결과표명：재허경솔위10－3시，분형방법대약기복화중등기복운배경하홍외소목표검측개솔체도1.0，대강기복운배경하홍외소목표검측개솔체도0.6，교현유기타일사방법구유경호적검측효과화배경괄응능력。
To solve the problem that the detection of infrared small targets under cloud background is strongly depend on the fluctuation of cloud,the fractal method is proposed. Firstly,the actual cloud background image is modeled by the fractional Brownian random field,and the fractal curve is obtained. Then the influence of small targets on the frac-tal characteristics of cloud background is analyzed,and the conclusion is draw that the fractal curve area will increase if small targets exist under cloud background. Finally,the approach to detect small targets is proposed,which is based on the fractal curve area dispersion. The results of experiments show that the target detection probability is up to 1.0 under weak and median cloud background and 0.6 under strong cloud background through fractal method when false alarm probability is 1 0 -3 ,which reflects the fractal method is more detective and adaptive to various cloud background than existing methods.