| The emerging of third generation mobile system (3G) makes various video applications in wireless environment possible. Original video data must be compressed for transmission in wireless network because of its high bandwidth. The newest video coding standard H.264/AVC developed by ITU-T and ISO, which has the characteristic of high compression rate and robust adaptability for different networks, was selected by 3GPP/3GPP2 as one of the video codec tools in 3G. Because of wireless network's characteristic of time variation, high bit error rate and limited bandwidth, so H.264/AVC using in 3G network has some special points.In this paper, firstly we introduced the basic video coding theories and video coding tools of H.264/AVC, and then we presented compressed H.264/AVC video transmission in 3G wireless network, together with the offline video transmission simulation in 3G wireless network. Secondly we studied the error resilience tools and error resilience schemes of H.264/AVC in 3G wireless network, here we found some basic principles for error resilience tools selection, and also presented some error resilience schemes for various video applications in 3G Thirdly we addressed two different rate control method for 3G, and we focused on the realization of a rate control method base on EBR technology. At last we introduced some application topics, such as H.264/AVC conformance, patent license and the comparison of H.264/AVC and AVS-P7.In this paper, we got some achievements listed as follows:a) A video communication simulation model was presented, which was suitable for bidirectional conversational video transmission simulation in 3G wireless network.After the birth of H.264/AVC video coding standard, VCEG experts have presented an offline model for video transmission simulation in 3G wireless network. It was widely used because of its simplicity and convenience for use. But the shortcomings of this model were obvious, because it was not suitable for use in bidirectional conversational video transmission simulation. Here a new model was presented, which used reverse channel of bidirectional transmission to transmit feedback information to video encoder, and then video encoder can make changes accordingly and improve error resilience ability of output video stream.This model can work under different time delay and different parameters, and its validity and ability was proved in latter experiment in this paper.b) Some error resilience selection principles for H.264/AVC encoder were presented for video transmission in 3G wireless network.Error resilience in encoder is important for all video applications, and H.264/AVC has provided many error resilience tools, but how to select and make full use of these tools doesn't have some basic standard. In this paper, we analyzed various error resilience tools and their applicability in 3G wireless network, and then we presented some error resilience tools selection principles in this situation. That is, in this error-prone and bandwidth-limited environment, we should not only consider error tools' rectification ability but also consider the limitation of bit rate, so we should make full use of spatial correlation of macroblocks to improve encoding efficiency, and complex FMO styles and random intra macroblocks refreshing were disadvantageous for video efficiency improvement.c) An implement scheme was presented for video rate control based on EBR technology of H.264/AVCTraditional VBR rate control method can smooth the output of the encoder in a relatively long time to reach an objective bit rate, so the output bit rate of each frame may vary according to its coding complexity. But EBR rate control method split one frame into fixed slices, and the output bit rate of each slice cannot exceed a fixed size, so the implement method of EBR rate control is different from the VBR rate control. This paper implemented an EBR rate control method based on dynamic slice structure. The first I frame and P frame in a GOP are encoded by a try-and-test method, then the following P frames in the GOP are encoded by a MAD predication method, here we use a line model to predict MAD value from previous P frame. Compared with simple slice method, our method can efficiently use channel resources and gets a better encoding quality.
|
PDF Full Text Request