中文说明：背景差分法又称背景减法，背景差分法的原理是将当前帧与背景图像进行差分来得到运动目标区域，这种方法较帧差法能更好的识别和提取运动目标，是目前运动分割中最常用的一种方法。但是需要构建一幅背景图像，这幅背景图像必须不含运动目标，并且应该能不断的更新来适应当前背景的变化，构建背景图像的方法有很多，比较常用的有基于单个高斯模型的背景构建，基于混合高斯模型的背景构建，基于中值滤波器的背景构造，基于卡尔曼滤波器的背景构造，基于核函数密度估计的背景模型构造。
English Description：
Background subtraction method is also called a background subtraction, background subtraction method is the difference to get moving with the background image of the current frame target range, the method can more frame difference method better identification and extraction of moving targets, is currently the most commonly used in motion segmentation method. But needs building a site background image, this site background image must not containing movement target, and should can constantly of update to adapted current background of changes, building background image of method has many, compared common of has based on single Gaussian model of background building, based on mixed Gaussian model of background building, based on in the value filter of background constructed, based on Kalman filter of background constructed, based on nuclear function density estimated of background model constructed.