9512.net
甜梦文库
当前位置:首页 >> >>

VIDEO CODING WITH SPATIO-TEMPORAL SCALABILITY FOR APPLICATIONS IN HETEROGENEOUS COMMUNICATI


VIDEO CODING WITH SPATIO-TEMPORAL SCALABILITY FOR APPLICATIONS IN HETEROGENEOUS COMMUNICATION NETWORKS Adam ?uczak, S?awomir Ma?kowiak, Marek Domański
Poznań University of Technology, Institute of Electronics and Telecommunications, Piotrowo 3A, 60-965, Poznań, POLAND, e-mail: {aluczak, smack, domanski}@et.put.poznan.pl
Abstract: The paper describes scalable hybrid video codecs for applications in heterogeneous communication networks characterized by various levels of Quality of Service. The proposed functionality of spatio-temporal scalability can be exploited for construction of highly scalable video codecs that produce several layers of bitstream thus allowing great flexibility of network management. The proposed hybrid solution exploits MPEG-like algorithm in P- and B-frames and wavelet decomposition in I-frames. The bitrate ratio of two consecutive layers can be about 1:2 up to 1:2.5. The paper comprises experimental results obtained for a two-layer system for progressive video sequence compression. Keywords: Hierarchical video coding, video data compression, MPEG, spatio-temporal scalability

1 INTRODUCTION
Recently, video and multimedia services are being rapidly developed. Their availability depends strongly on communication network infrastructure. High bitrates needed for video transmission impose severe requirements on communication networks. Existing communication networks are very inhomogeneous. There exist links of various available bitrate and various error rate (Fig.1). Therefore there exist growing interests in video transmission through heterogeneous communication networks that are characterized by various available levels of Quality of Service.

Different levels of Quality of Service (QoS) are often related to different available transmission bitrates. On the other hand, the service providers demand that the data are broadcasted once to a group of users accessed via heterogeneous links. For this purpose, the transmitted bitstream should be partitioned into some layers in such a way that lower layers can be decoded independently from the others thus producing images with reduced quality (progressive coding) or resolution (hierarchical coding). Various layers 1,...,N correspond to subsystems N1 to NN of a heterogeneous communication network (Fig. 2).

Modem PTSN Video Service

ATM

ADSL

GSM

ISDN

Figure 1. A heterogeneous communication network with video services.

Digital video input

Scalable video coder

Layer N . . . Layer 3 Layer 2 Layer 1 N1 N2

. . .

NN

N3

Quality of service

Figure 2. Multi-layer scalable video coding system in heterogeneous network that consists of sub-networks N1,...,NN. In this paper, a structure of a hybrid highly scalable video coder is proposed. The goal is to achieve total bitrate of all layers of scalable coding possibly close to the bitrate of single-layer coding. The assumption is that high level of compatibility with the MPEG video coding standards would be ensured. In particular, it is assumed that the low-resolution base layer bitstream is fully compatible with the MPEG-2 standard which describes scalability with low numbers of available layers corresponding to various levels of QoS [1].

ω2

7

8

4 1 0 0 2 3

5 9 6 ω1

2 INTRAFRAME SCALABLE CODING
The intraframe spatially scalable (hierarchical) coding can be implemented using wavelet decomposition as it was already proposed in quit many references [2-6]. The layers correspond to various resolution images compressed. Figure 3 illustrates relations between individual layers and and spatial frequency bands of the images encoded on a given layer.

Figure 4. Subbands obtained using wavelet decomposition. The layers correspond to subbands or groups of subbands obtained using wavelet decomposition (Fig.4). The correspondence between subbands and layers is shown in Table 1. Table 1. Subbands and the corresponding layers by intraframe coding. Layer Subbands L1 0 L2 1,2,3 L3 4,5,6 L4 7,8,9

ω2 L4

L3
L2

3 INTERFRAME VIDEO CODING WITH THE FUNCTIONALITY OF SPATIO-TEMPORAL SCALABILITY
Unfortunately, spatially scalable systems proposed by MPEG-2 coding standard are inefficient because the bitrate overhead is very high as compared to bitrate of a single-layer MPEG encoder. Recently proposed MPEG-4 has proposed no substantial improvement. Therefore there were many attempts to improve the scheme of

L1 0 ω1

Figure 3. Layers in the spatial frequency domain.

spatial scalability by application of subband decomposition [2-6]. Unfortunately, this approach suffers from problems with right bit allocation to the layers. In order to avoid the above mentioned problem, spatio-temporal scalability is proposed [7,8]. Here, a low layer corresponds to pictures with reduced both spatial and temporal resolution. Higher layers are used to transmit the information needed for restoration of higher spatial and temporal resolution. Temporal resolution reduction is achieved by partitioning of the stream of B-frames: each secIn DCT

ond frame is included into the enhancement layer only. Therefore there exist two types of B-frames: ? BE-frames which exist in the enhancement layer only and ? BR-frames which exist both in the base and enhancement layers. The structure of the proposed coder in the interframe mode of operation is given in Fig. 5. In fact, it is slightly modified structure of a coder with the functionality of spatial scalability as proposed in the MPEG-2 standard [1].

Q IQ IDCT

VLE

bits_e

mv

MCP

FS

ME Spatial Enhancement Encoder Spatial Decimator Spatial Interpolator

Motion Compensated DCT Encoder Base Encoder

bits_b

DCT Q IQ IDCT

Discrete cosine transform Quantization Dequantization Inverse cosine transform

FS MCP ME

Frame memory Motion-compensated predictor Motion estimator

Figure 5. Interframe coder: the two-layer case.

4 EXPERIMENTAL RESULTS
In order to verify the structure proposed a verification model have been prepared. The software is currently available for progressive sequences with the input resolution defined by the standard digital television resolution. The experiments have been made with progressive 720 × 576, 50 Hz, 4:2:0 test sequences (cf. Table 2). Frames from the test sequences are shown in Figs. 6 and 7.

For a 16-frame GOP, the structure of the bitstream is shown in Fig. 8. It is assumed that a GOP consists of 12 B-frames, 3 P-frames and 1 I-frame. The obtained results prove that the structure proposed can be used for multi-layer systems where the bitrate ratio of two consecutive layers is about 1:2 up to 1:2.5.

Table 2. The experimental results. Funfair Bitstream [Mb] Average PSNR [dB] for luminance Base layer bitstream [Mb] Scalability overhead [%] 5.63 32.17 2.16 8.7 Flower Garden 6.40 30.92 2.17 21.5 Stefan 5.85 35.11 2.14 13.8 Cheer 5.35 31.99 2.15 2.7 Bus 6.33 34.51 2.15 22.0

BR

I

Base layer
BE P

B

Enhancement layer

P

I

Figure 8. A typical bitstream structure.

REFERENCES
[1] B. Haskell, A. Puri, A. N. Netravali, Digital Video: An Introduction to MPEG-2, New York: Chapman & Hall, 1997. T. Tsunashima, J. Stampleman, V. Bove, A scalable motion -compensated subband image coder, IEEE Trans. on Communication, vol. 42, pp. 1894-1901, 1994. F. Bosveld, Hierarchical video compression using SBC, Ph.D. dissertation, Delft University of Technology, Delft 1996. H.Gharavi, W.Y.Ng, H.263 Compatible Video Coding and Transmission, in Proc. First International Workshop on Wireless Image/Video Communication, pp. 115-120, Loughborough 1996. Senbel, H. Abdel-Wahab, Scalable and robust image compression using quadtrees, Signal Processing: Image Communication, vol. 14, pp. 425-441, 1999. Shen, E. Delp, Wavelet based rate scalable video compression, IEEE Trans. Circuits Syst. Video Technology, vol. 9, pp. 109-122, February 1999. M. Domański, A. ?uczak, S. Ma?kowiak, R. ?wierczyński, Hybrid coding of video with spatio-temporal scalability using subband decomposition, in Signal Processing IX: Theories and Applications, pp. 53-56, Rhodes, September 1998. M. Domański, A. ?uczak, S. Ma?kowiak, R. ?wierczyński, U. Benzler, Spatio-temporal scalable video codecs with MPEG-compatible base layer, in Proc. Picture Coding Symposium 1999, pp. 45-48.

Figure 6. A frame from the test sequence Cheer.

[2]

[3] [4]

[5]

[6]

[7]

Figure 6. A frame from the test sequence Funfair.

5 CONCLUSIONS
It is shown that application of a hybrid scheme that combines wavelet decomposition with MPEG-like structure of interframe coding leads to promising results which could be exploited by construction of multi-layer systems.
[8]


赞助商链接

更多相关文章:
更多相关标签:

All rights reserved Powered by 甜梦文库 9512.net

copyright ©right 2010-2021。
甜梦文库内容来自网络,如有侵犯请联系客服。zhit325@126.com|网站地图