基于背景建模和前景建模的红外运动检测

Background and Foreground Modeling-based Infrared Motion Detection

为抑制了红外图像信噪比低、非静止背景等引起的虚警产生,将基于背景建模和前景建模运动检测(BM-FM)算法引入红外面目标检测中;为抑制因目标与背景对比度低引起的部分目标区域漏检,提出了基于空间邻域信息的前景建模方法(SNFM),利用前一帧图像在像素点邻域内的前景目标点为样本,用核密度估计进行前景建模。试验表明了基于该前景建模方法的BM-FM算法的有效性。

The background and foreground modeling detection method （BM-FM）, which is used in infrared moving object detection, can effectively restrains false-positive caused by low signal-noise ratio, nonstationary background. For suppressing false-negative in object field where the intensity is similar with the background, propose a foreground modeling method based on spatial neighborhood information （SNFM）, which models the foreground probability for one pixel using pixels which are considered as object points in last frame in the neighborhood of this pixel. The result shows the validity of the BM-FM algorithm based on SNFM.