Yuzhuo Zhong

Performance and complexity joint optimization for H.264 video coding

This paper proposes a video coding optimization approach for those real-time visual communication applications with performance and complexity tradeoff based on the coming H.264 video coding standard, which has a very high performance with high coding complexity. In H.264, the better coding performance is achieved at the cost of complexity, and it is too complex to be applied widely. With the performance and complexity joint optimization method of this paper, the modules are optimized respectively in the order of their computation proportion. Adaptive hexagon-based search (AHBS) and memory motion compensation (MMC) are proposed for encoder and decoder. AHBS is about 40 times faster than full search, and a little faster than diamond search. However AHBS achieves the best coding performance: 0.4dB improvement in low motion sequences, and 0.2dB improvement in high motion sequences compared with full search. The decoding process is accelerated up to 5 times totally with proposed MMC and other optimized techniques. Therefore the efficiency of codec system is improved greatly. Optimized video codec can be used in uni-casting and broadcasting streaming applications with limited computation resource in those embedded movable devices.

Position prediction motion-compensated interpolation for frame rate up-conversion using temporal modeling

In this paper, a new motion-compensated interpolation method (MCI) for frame rate up-conversion is proposed. In conventional MCI algorithms using transmitted motion vectors (MVs), the block artifacts are caused by the unreliable MVs from the encoder. In our proposed method, motion vectors correction (MVC) is applied on the transmitted MVs before interpolation considering the spatial correlation. In consideration of the temporal correlation, generative model is used to generate the temporal-predicted MVs. Then position prediction method is applied on the transmitted MVs and the temporal-predicted MVs separately to predict the positions the interpolated blocks really move to, which makes the MVs used for interpolation nearer to the true motion. The final MVs are selected from the two candidates of position prediction by the minimum sum of absolute difference (SAD). Applied to the H.264 decoder, our proposed method achieves significant increase on PSNR and obvious decrease of the block artifacts.

A novel video coding framework by perceptual representation and macroblock-based matching pursuit algorithm

This paper presents a novel hybrid video coding framework by perceptual representation and macroblock-based matching pursuit algorithm (PRMBMP), which uses a set of filters to extract the perceptual parts of each video frame. The original video frame is separated into low-frequency image and high-frequency image. The low-frequency image has low sensitivity to human perception and few complex texture details, which can be handled efficiently by traditional H.264 video coding. The high-frequency image is the perceptual representation, which includes more texture details and edges. The proposed macroblock-based matching pursuit algorithm is used to compress the high-frequency image, which speeds up the conventional matching pursuit algorithm efficiently by estimating the local optimization. The experiments show that the proposed framework can achieve 2 to 3 dB improvements compared with the conventional H.264 video coding framework. The proposed framework also has the definition scalability, which can be widely used in bandwidth-variation video applications.

Characterizing user behavior to improve quality of streaming service over p2p networks

The universal recognition that it is critical to improve the performance of existing systems and protocols with the understanding to practical service experiences motivates us to discuss this issue in the context of peer-to-peer (P2P) streaming. With the benefit of both practical traces from traditional client-server (C/S) service systems and logs from P2P live broadcasting system, in this paper we first characterize end user behaviors in terms of online duration and reveal the statistically positive correlation between elapsed online duration and expected remaining online time. Then we explore the feasibility to improve the quality of streaming service over P2P networks by proposing Low Disruption Tree Construction (LDTC) algorithm to take the online duration information into account when peers self-organize into the service overlay. The experiment results show that LDTC could achieve higher stability of video date delivery tree and in turn improve the quality of streaming service.

Improving quality of live streaming service over p2p networks with user behavior model

In this paper, we mainly investigate how to improve the streaming quality of service over Peer-to-Peer (P2P) networks with the benefit of practical service traces. Considering that the QoS of P2P live streaming is generally affected by high dynamics in peer community, we abstract the online duration model of end users and then exploit this characteristic to improve the stability and encourage contribution in P2P networks, which in turn facilitates the quality of streaming. The first effort involves the online information when constructing video delivery tree and hence achieves less playback disruption. The other algorithm takes the online duration into account when selecting neighbor peers for data exchange and advocates more mutual cooperation between peers. The experiment results show that the proposed schemes could respectively achieve higher stability and better delivery quality.

A method of keyframe setting in video coding: fast adaptive dynamic keyframe selecting

In this paper, we describe a novel method of keyframe setting in video coding: fast adaptive keyframe selecting. The method analyzes video contents and decides keyframe setting dynamically and adaptively. Utilizing information generated by motion-estimation, our method involves only a little additional computing cost. Our method can improve both objective and subjective quality compared with traditional keyframe setting. It can be applied to any motion-estimation based video compression. We tested the method on not only standard test sequences but also real movie sequences proving that it gives valuable results.

Bilock-based fine granularity scalable video coding for content-aware streaming

Video streaming is becoming more and more popular with widely used hybrid networks. The prime challenge of such applications is to deal with varying transmission bandwidth. This paper proposes a block-based fine granularity scalable (FGS) coding structure, which is a more flexible scalable video coding structure supporting content-aware streaming compared with the MPEG-4 FGS coding structure. The streaming server can conveniently implement content-aware rate allocation or content-based selective enhancement dynamically through users interaction with the proposed scalable coding structure. Thus the streaming server can have a differentiated delivery strategy according to users preference. However the uniform rate allocation for bit-stream truncation of the proposed block-based FGS will result in more than 2dB loss by PSNR compared with MPEG-4 FGS within quite a wide range of bit rates. A fast optimal rate allocation method is also proposed to solve this problem in this paper. The coding efficiency is improved, which can be comparable with MPEG-4 FGS coding and is even better (0.5dB) with some sequences at some bit rates.

Optimal caching algorithm based on dynamic programming

With the dramatic growth of multimedia streams, the efficient distribution of stored videos has become a major concern. There are two basic caching strategies: the whole caching strategy and the caching strategy based on layered encoded video, the latter can satisfy the requirement of the highly heterogeneous access to the Internet. Conventional caching strategies assign each object a cache gain by calculating popularity or density popularity, and determine which videos and which layers should be cached. In this paper, we first investigate the delivery model of stored video based on proxy, and propose two novel caching algorithms, DPLayer (for layered encoded caching scheme) and DPWhole (for whole caching scheme) for multimedia proxy caching. The two algorithms are based on the resource allocation model of dynamic programming to select the optimal subset of objects to be cached in proxy. Simulation proved that our algorithms achieve better performance than other existing schemes. We also analyze the computational complexity and space complexity of the algorithms, and introduce a regulative parameter to compress the states space of the dynamic programming problem and reduce the complexity of algorithms.

Joint Inter and Intra Shot Modeling for Spectral Video Shot Clustering

This paper proposed a novel video shot clustering algorithm using spectral method by joint modeling of inter and intra shot. Gauss mixture model (GMM) is used for probabilistic space-time modeling of intra-shot pixels. The spectral clustering method is applied on the GMM parameters. The problem of automatic model selection is currently an open issue for spectral method. Here we propose a novel automatic model selection based on joint inter-intra model optimization to achieve the global optimization in both model parameters and cluster numbers. We compare the proposed method with the conventional spectral method for sports video clustering. The simulation results show more accuracy on how many clusters and clustering results

TPOD: a trust-based incentive mechanism for peer-to-peer live broadcasting

The prevalence of emerging peer-to-peer (P2P) live broadcasting applications has practically demonstrated that they could scale to reliably support a large population of end users. However, these systems potentially suffer from two major threats: peers generally interact with unfamiliar partners without the benefit of trusted third party or verification authority, resulting in poor service if meeting with unreliable upstream nodes, while peers essentially tend to be selfish when it comes to the duty rather than the benefits and hence undermine the system performance. The trust and cooperation issues motivate us to investigate the design of trust-based incentive mechanism which establishes trustful relationship among peers and balances what they take from the system with what they contribute. The proposed TPOD mechanism leverages the statistical analysis to the practical service logs of client-server and P2P systems and effectively offers incentive through service differentiation. It goes beyond existing approaches in the following four desirable properties: (1)Trust-based; (2)Practical-oriented; (3)Objective metrics about past behaviors and (4)Distributed nature upon gossip-based overlay. The experiment results over PlanetLab verify its effectiveness.