Real Time  Scalable Bitstream Extraction
2005 . 03 ~ 2006. 02
1. Introduction
 Video streaming service is widely used according to the development of wired and wireless network environment and video coding technology. The streaming video service in the fluctuant and heterogeneous network environments is required by ubiquitous video consumption. To satisfy these requirements in the streaming video service, we can use the SVC (MPEG-4 Scalable Video Coding) bitstream whose syntax enables a scalable data extraction. Further, it is necessary to extract a bitstream in real-time by changing scalabilities.

In our research, we propose the model of real-time bitstream extractor. And we also suggest the extraction methods of the bitstream with changing scalability in real-time.


2. Real time Scalable Bitstream Extraction
 For QoS guaranteed video streaming in the fluctuated network, one needs to change bitstream scalability in real-time. In this project, we propose a real-time bitstream extractor. To have the real-time bitstream extractor, the following three points could be considered. One is that bitstream information for the extraction is analyzed and stored before actual extraction process. Second, the extractor should have appropriate structure to be able to process in real-time according to varying input parameters. Third, the extractor should have a reasonable processing unit to extract the bitstream in real-time by changing scalability. Figure 1 shows the proposed structure (a) and algorithm (b) of the bitstream extractor.

                        Fig. 1.      (a) Extractor Structure                                                        (b) Extractor Algorithm
 As shown in figure 1, the proposed extractor consists of two parts: Bitstream Analyzer, Bitstream Extractor. Input parameters of the proposed extractor have four values: resolution, frame rate, required bitrate, and CGS_Flag. Input parameters decide a scalability measure for the extracted bitstream. The processing unit which works in real-time is a GOP unit, which is minimum data unit to support the scalability of SVC bitstream. The bitstream analyzer analyzes and keeps the bitstream information needed for the extraction. The process of extraction is performed in the bitstream extraction part.

 Fig. 2. Extraction point decision module.


The proposed extraction algorithm in this project is composed of 3 steps. First step is to decide a proper extraction point by using input parameters and bitstream information. The extraction point decision module decides the decision of extraction point and outputs 3 parameters: Ext_Spatial_Layer, Ext_Temporal_Level, Ext_FGS_Layer. The Ext_Spatial_Layer has the value of dependency id of the input resolution. The Ext_Temporal_Level has the temporal level of the input frame rate. The Ext_FGS _Layer is decided by the required bitrate. In second step, input source bitstream is divided into GOP unit in the bitstream separation module and the segmented bitstream is sent to the extraction module. In third step, extraction module extracts the segmented bitstream by using the output parameter of the extraction point decision module(figure 2). Three steps are repeated until the end of input bitstream.



Fig. 3. Temporal(Frame rate) Change


Fig. 4. Quality(SNR) Change


3.  Experiments
 The information of test SVC bitstream for the experiment is shown below(Table1, 2 and 3).

Table 1.Test Bitstream Information


Table 2.Bitstream Information


Table 3.Input Parameter


 Follows figures are Implemented GUIs.

Fig. 5.Server GUI


Fig. 6.Client GUI

 Below figure is Result of this Experiment.

This result is something for validating the proposed methods. Test result give the evidence for good working of proposed extractor with proposed extraction method. As a result, we identify the validation of the proposed methods in this project.

Fig. 7. Result

배태면, 김덕연, 노용만, 강정원, 김재곤,”실시간 스케일러빌러티 변환 SVC 비트스트림에 대한 연구”, 방송공학회 추계학술대회
배태면, 김영석, 노용만, 최해철, 김재곤, "QoS 기반 스트리밍 서비스를 위한 실시간 SVC 비트스트림 추출기에 대한 연구", 전자공학회 추계학술대회.
Tae Meon Bae, Duck Yeon Kim, Yong Man Ro, Hae Chul Choi, Jae Gon Kim ,"Real-time Scalable Bitstream Extraction in Video Streaming Service" WIAMIS 2006 (submitted)
Dong San Jun, Jung Won Kang , Hae Chul Choi, Jae Gon Kim,Tae Meon Bae, Yong Man Ro, “Dynamic spatial layer switching support at the point of IDR picture” JVT-R060, 75th MPEG JVT meeting

View The Demo