基于帧间扰动相关性的弱小目标提取方法

Extraction Method of Weak and Small Target Based on Correlation of Inner-Frame Disturbance

帧差分和背景差分技术在运动目标检测方面具有很好的原理性效果;但是对弱小目标图像来说,当目标尺寸和噪声、干扰点尺寸能够比拟,硬件系统和信号噪声的随机性会使差分方法几乎完全失效;在研究差分结果噪声相关性的基础上,提出定点能量积累,避免了能量积累时对目标运动速度的限制;并利用组合扰动排除法,大幅度削弱了帧间相互独立的噪声和杂波干扰;实验结果表明,该方法能够以较低的运算量准确地检测出强噪声干扰和复杂背景下的运动弱小目标,实时性好。

There is good effect in principles in moving target detection by using Frame difference and background difference. However, as to weak-- and-- small target images, randomicity of noises which exist in hardware system and signals will almost invalidate the method of difference when the size of targets is very similar to that of noises and disturbance points. On the foundation of researching the noises correlation among difference results, energy accumulation of immovable points is proposed to avoid the speed limit of moving target while accumulating energy. And by means of combinating disturbance elimation, absolute noises and clutter among frames are weakened greatly. The results of experiment show that this method can detect moving weak--and--small target covered under high noise disturbance and complex background by less computation, and it is of good real--time performance.