Nonlinear sampling control model for real-time video sequences of compressed domain

For the reliable video transmission in bandwidth-limited overlay multicast network (OMN), a nonlinear sampling control model is proposed to extract the key frames in video packet sequence of compressed domain according to the characteristics of video packets transmitted in OMN. In the model, weighted membership function and discriminating rules are defined, and key frames are sampled nonlinearly among compressed video sequences on the basis of semantic information of frames in compressed domain. Results of theory analysis and practical test show that video sequences are abstracted by decreasing the frame rates without the loss of spatial resolution to meet the demand of available bandwidth of transmitting the video sequences.