背景差分法是采用图像序列中的当前帧和背景参考模型比较来检测运动物体的一种方法,其性能依赖于所使用的背景建模技术。背景差分法检测运动目标速度快，检测准确，易于实现，其关键是背景图像的获取。在实际应用中，静止背景是不易直接获得的，同时，由于背景图像的动态变化，需要通过视频序列的帧间信息来估计和恢复背景，即背景重建，所以要选择性的更新背景。-Background difference method is the use of images in the sequence of the current frame and the background reference model is a method to detect the movement of objects, its performance depends on background modeling technology used. Background subtraction method to detect moving objects fast, accurate detection, easy to implement, the key is to get the background image. In practice, the background is still difficult to obtain directly, while, due to the dynamic changes of the background image, through inter video sequence information to estimate the background and restore that background reconstruction, so to selectively update the background.