Abstract: Through an analysis of the structure and characteristics of video data, the paper describes methods to realize four important modules of the video coding system: scene analysis, shape coding, motion coding and texture coding. It describes the data structure of important objects in this system. In particular it introduces some new methods adopted in all modules: the method of automatic extraction of moving objects using multiple features and multiple frames, the technology of image filling and method of discrete wavelet transform. A content based video coding system is finally implemented with a complete set of functions.