VBR video traffic modeling based on scene change detections using slice-level traffic characterizations = 슬라이스 단위 트래픽 분석 및 장면 전환검출에 의한 가변전송율 비디오 트래픽 모형화

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 518
  • Download : 0
The asynchronous transfer mode (ATM) networks are expected to support diverse services with a wide range of traffic characteristics. In particular, the statistical characterization of coded video traffic generated by variable bit rate (VBR) encoders is very important for the design of traffic control mechanisms in ATM networks. However, the characterization of this traffic has been very limited. In this dissertation, we characterize VBR video traffic based on the actually measured data. First, we characterize JPEG-like and H.261-like VBR traffic at slice level without scene changes by introducing vector autoregressive processes (VARs). We show that slice-level video traffic sources generally exhibit periodic mean variations and that residual slice-level bit-rates also greatly vary depending on their positions in a picture. The periodic mean results from the spatial nonstationarity which is a substantial nature of input videos. In particular, the JPEG-like slice-level bit-rate sequences with low activities are approximately deterministic and periodic. On the other hand, one slice in the previous picture only tends to affect the residual bit-rate of the corresponding slice in the current picture. By decomposing overall autocovariance functions into their periodic mean and residual components, we also show that the VAR-based autocovariance functions can fit the measured ones very well. Next, by utilizing these characterization results, we propose three systematic scene change detection methods from JPEG-like slice-level bit-rate sequences: two methods based on abrupt changes of the sample correlation coefficients (SCCs) and one improved method of the conventional methods based on abrupt changes of the picture-level bit-rates. We also propose an SCC-based method for identifying pictures involving flashes or noises; there have been few previous works on identifying these pictures. We can obtain the SCCs by the regression analysis between the bit-rates of successive ...
Kim, Jae-KyoonSung, Dan-Keun김재균성단근
한국과학기술원 : 전기및전자공학과,
Issue Date
101736/325007 / 000875219

학위논문(박사) - 한국과학기술원 : 전기및전자공학과, 1995.8, [ viii, 157 p. ]


Scene Change; ATM; MPEG; 비디오; 트래픽 모형화; 장면 전환; 표본상관계수; 슬라이스; Video; Traffic Modeling

Appears in Collection
Files in This Item
There are no files associated with this item.


  • mendeley


rss_1.0 rss_2.0 atom_1.0