针对摄像机固定下的复杂背景环境，对采集到的视频图像的图像数据用混合高斯背景建模方法实现前景/背景分割，实现运动目标检测和跟踪。在进行前景检测前，先对背景进行训练，对图像中每个背景采用一个混合高斯模型进行模拟，每个背景的混合高斯的个数可以自适应。然后在测试阶段，对新来的像素进行GMM匹配，如果该像素值能够匹配其中一个高斯，则认为是背景，否则认为是前景。由于整个过程GMM模型在不断更新学习中，所以对动态背景有一定的鲁棒性。最后通过对一个有树枝摇摆的动态背景进行前景检测，取得了较好的效果。-For complex background environment under fixed camera, the image data captured video images using Gaussian mixture background modeling method implementation foreground/background segmentation, detecting and tracking moving objects. Before foreground detection, the first training background, the background of each image using a Gaussian mixture model to simulate the number of each Gaussian mixture background adaptively. Then in the testing phase, the pixels on the new GMM match, if the pixel value is able to match one of the Gaussian, then that is the background, otherwise considered a prospect. Since the whole process is continuously updated GMM model in the study, so the dynamic context of a certain robustness. Finally, on a tree swing dynamic background foreground detection, achieved good results.