Catálogo de publicaciones - libros

Flexible macroblock ordering (FMO) is an error resilience feature of H.264 for video transmission over packet-lossy networks. Error concealment (EC) provides a basic error resilience tool for decoder to recover lost information and to reduce error propagation effect. This paper presents an analysis on the dependency of EC performance on FMO modes through the investigation of the expected number of correctly received neighboring macroblocks for a lost macroblock. We present simulation results and performance evaluation of FMO with different encoding parameters in various packet loss rates in the context of EC performance. Simulation results show that FMO provides an effective feature for PSNR improvement in environments with high packet loss rates, especially when intra-frame period is large.

A visual measure for the purpose of video compressions is proposed in this paper. The novelty of the proposed scheme relies on combining three human perception models: motion attention model, eye movement based spatiotemporal visual sensitivity function, and visual masking model. With the aid of spatiotemporal visual sensitivity function, the visual sensitivities to DCT coefficients on less attended macroblocks are evaluated. The spatiotemporal distortion masking measures at macroblock level are then estimated based on the visual masking thresholds of the DCT coefficients with low sensitivities. Accordingly, macroblocks that can hide more distortions are assigned larger quantization parameters. Experiments conducted on the basis of H.264 demonstrate that this scheme effectively improves coding efficiency without picture quality degradation.

In this paper, we propose a hybrid deblocking algorithm to improve the visual quality of block-based low bit rate coded images, those are processed both spatial and wavelet domains. The proposed algorithm reduces the blocking artifacts using the statistical characteristics of block discontinuity as well as the lipschitz regularity along the behavior of wavelet coefficients across scales. In this algorithm, detection of blocking artifacts and block boundary classification is performed in spatial domain to reduce the computational complexity for performing the wavelet transform and inverse wavelet transform for all image, and adaptive filtering is processed in spatial or wavelet domains. Spatial adaptive filtering is processed to reduce the blocking artifacts in smooth region with the blocking artifacts. For complex or mixed region, the Lipschitz regularity is obtained to analyze the evolution of the local maxima of the wavelet transform modulus across scales and these irregular singularities are adaptively removed in wavelet domain. Experimental results show that the proposed algorithm produced better results than those of conventional algorithms both PSNR and visual quality.

In in-band wavelet video coding schemes, motion prediction is applied in the spatial subband domain. Compared to motion prediction in full-resolution image domain, in-band schemes suffer coding performance loss at full resolution. One reason is that signals of the subband at low resolution are predicted from the reference frames at low resolution, which has comparatively low quality. However, if signals of the subband at high resolution are involved in the prediction of signals at low resolution, mismatch will occur in decoding low-resolution video when the corresponding signals of high resolution are not available at the decoder. This paper first analyzes the mismatch error propagation when low-resolution video is decoded. Then based on the analysis we propose a frame-based cross-resolution leaky prediction scheme for in-band wavelet video coding to make a good trade-off between reducing mismatch error of low resolution and improving coding performance of high resolution. Experimental results show that, the proposed scheme can dramatically reduce the mismatch error by about 0.3~2.5dB at different bit rates for the low resolution, while for the high resolution, the performance loss is marginal.

In this paper, we propose an efficient 4×4 intra prediction mode decision method for the complexity reduction of H.264 video coding. Based on the observation that a good prediction could be achieved when local edge direction of a block is in the same direction as the intra prediction mode, local edge direction of block is obtained in transform domain to filter out majority of intra prediction modes. By filtering out majority of intra prediction modes, we only have to consider a candidate set of most highly probable 4×4 intra prediction modes for which the rate distortion optimization process should be performed. Experimental results show that the proposed method can achieve a considerable reduction of computation while maintaining similar PSNR and bit rate.

In the joint scalable video model (JSVM), selection of quantization parameters for mode decision (QPMD) and bit rate control (QPBC) is important for efficient coding performance. For quality (SNR) scalability, QPMD is adjusted only to the base layer QPBC in JSVM. Thus, it reduces coding efficiency in the enhancement layer. In this paper, we propose a new method for selecting optimum quantization parameters for mode decision (OQPMD) in order to improve coding efficiency for both the base and the enhancement layers. For the base layer, we propose optimum scaling factors in each decomposition stage. We also propose an offset quantization parameter for the enhancement layer. Experimental results show that the proposed method increases the average PSNR value up to 0.8dB.

In this paper, we design the metadata model for Event notification that can be utilized for various commercial purposes on interactive broadcasting environment, and implement the testbed system based on the metadata model. The metadata model is based on the ISO/IEC 21000-15 and we extended some descriptions which have been adopted as ISO/IEC 21000-15 Committe Draft. Event notification service based on this metadata model achieves a step towards the healthy commercial future of interactive service by enabling target marketing, the audience rating analysis, user preference analysis, and monitoring illegal copies.

Due to the limitation of broadcasting service, in general, TV programs with commercial advertisements are scheduled to be broadcasted by demographics. The uniformly provided commercial can not draw many TV viewers’ interest, which is not correspondent to the goal of the commercial. In order to solve the problem, a novel target advertisement technique is proposed in this paper. The target advertisement is a personalized advertisement according to TV viewers’ profile such as their age, gender, occupation, etc. However, viewers are usually reluctant to inform their profile to the TV program provider or the advertisement company because their information can be used on some bad purpose by unknown people. Our target advertisement technique estimates a viewer’s profile using Normalized Distance Sum and Inner product method. In the experiment, our method is evaluated for estimating the TV viewers’ profile using TV usage history provided by AC Neilson Korea.

To support personalized broadcasting service and T-learning, we propose to use TV-Anytime standard. The TV-Anytime Forum is an association of organizations which seeks to develop specifications to enable audio-visual and other services based on mass-market high volume digital storage in consumer platforms such as a set-top box (STB). TV-Anytime standard is classified to Phase 1 and 2 according to its functionality. The former supports PVR-based applications in a uni-directional broadcast and the latter deals with the sharing and distribution of rich content among local storage devices and/or network digital recorders in home network environments. In this paper, we first overview TV-Anytime specifications, and secondly present an end-to-end prototype system for personalized broadcasting services based on TV-Anytime Phase 1 standard which consists of metadata authoring tool, transmission server, and set-top box. At the last, we propose the method to utilize content package in TV-Anytime Phase 2 to develop a T-learning environment.

This paper describes SceneCabinet/Live!, which is a system that generates real-time segment metadata using audio/video indexing and natural language processing to provide viewers with a continuously metadata update service of live programs. It significantly reduces the costs of generating segment metadata that is relevant to the viewer and reduces the time by about half. It enables metadata to be sent without delay in a metadata updateable service. We approached using audio/video recognition technologies to make index of the scene like a book because video programs are difficult to search scenes in. Segment metadata is used on broadcasting-telecommunication linkage services, affording a new way to view programs on optical and broadcast networks by supporting TV digests and videos from VoD services. These new metadata services are more convenient to use and more enjoyable to watch than conventional ones.