对模型基编码方法进行了综述 ,着重叙述了目前模型基视频图像编码方法所遇到的问题和解决方法 ,其中许多问题都是我们目前正在研究的内容 ,包括 :3D线框模型、人脸检测、人脸纹理编码、运动参数估计和人脸模型参数编码等 .这些问题是模型基方法在实用前必须解决的问题 ,例如 ,如何在一幅图像中检测和定位人脸 ,如何从构成人脸模型的图像中提取参数 ,及如何进行运动参数的压缩等 .采用这种技术 ,讲话人的脸部图像可用一个模型来表示 ,它在各帧之间的运动可以用很少的参数来编码 ,从而在视频会议系统中使用模型基视频图像编码技术可以得到最大的视频图像压缩率 This paper summarizes the model-based coding methods and focuses on the problems and solutions of the current model-based video coding methods. Many of the problems are currently under study, including 3D wireframe models, face detection , Face texture coding, motion parameter estimation and face model parameter coding, etc. These questions are the problems that the model-based method must solve before practical application, for example, how to detect and locate human faces in an image, Extract the parameters of the model image and how to compress the motion parameters, etc. Using this technique, the speaker’s facial image can be represented by a model whose motion between frames can be encoded with a small number of parameters, So video conferencing system using the model based video coding technology can get the maximum video image compression rate