New techniques for streaming MPEG video over the internet
Many video coding standards have been established for various streaming video applications, such as H.263 for low-bit-rate two-way video communications, MPEG-2 for broadcasting and general high-quality video applications, and MPEG-4 for streaming video and interactive multimedia applications. Since its introduction a decade ago, streaming video over the Internet has been experiencing dramatic growth with the rapid increase of network bandwidth and computing power. The target of the streaming video service is to provide satisfactory visual quality and flexible functionality to the clients. Many issues should be taken into consideration, for example, how to deliver pre-encoded video streams over networks with varying bandwidth, how to improve the error resilience of the video transport over lossy channels, and how to support friendly user-interface (such as providing full VCR functionalities for controlling the playback of the video streams) at the client side.In this dissertation, we propose several new techniques to address the above issues. In the first part, we present an enhancement layer truncation scheme for the delivery of the MPEG-4 Fine-Granularity-Scalability (FGS) video. Our target is to reduce the quality variation of different parts within each frame when the last transmitted enhancement layer is truncated according to the available network bandwidth. In our approach, the bit-budget for the truncated enhancement layer is redistributed to each block, so that the whole frame area can be covered uniformly. An operational rate-distortion optimization scheme is further adopted to improve the decoded visual quality and to reduce the intra-frame quality variation. In the second part, we point out that multi-path feature exists in today's networks, and we can make use of it to overcome the problems of instantaneous network congestion and large bandwidth requirement for higher enhancement layers by transmitting the split FGS enhancement layers. We initially split the original enhancement layers evenly into multiple descriptions. Rate-distortion optimization is then carried out on each description to improve the performance. In the third part, we present our solution to provide full VCR functionality for the MPEG video streaming application by adding a reverse-encoded bit-stream and using a minimum-cost frame-selection scheme to minimize the number of frames to be sent over the network and to be decoded. The drift-compensation issue is also taken into consideration. With the proposed scheme, an MPEG video streaming system with full VCR functionality can be implemented to minimize the required network bandwidth and decoder complexity.
- Electrical engineering