Key Technologies In The Network Television

With the fast development of home digital entertainment and Internet techniques, great changes have already occurred in the transmission of TV programs and the way by which people watch TV programs. Integrating network techniques and digital video techniques, IPTV will not only provide us the traditional way to enjoy TV programs, i.e. watching TV programs by choosing channel (PUSH application, or broadcast application), but also provide us an active way to enjoy TV programs, i.e. watching TV program according to our interest (PULL application, or demand application). IPTV includes many key techniques, such as video streaming, networks multicast, video content analysis etc, which are the foundation that enable IPTV to be popular.Taking the two basic application ways of IPTV, i.e. broadcast application and demand application as the clues, we discuss some key techniques in this thesis. Based on the research on video content analysis and video streaming, this thesis puts much effort on some key techniques such as shot boundary detection, motion analysis in compressed video, scalable video coding, error concealment as well as video transcoding, and proposes some novel ideals or methods. The main work are summarized as follows:1. A fast algorithm to detect dissolve shot boundaries in compressed domain is proposed in the thesis: Among many different kinds of shot boundaries, dissolve is the most difficult to detect because it is the mix of two neighbor shots in temporal and spatial domain. We address an effective and fast algorithm to detect dissolve shot boundaries. Firstly it makes use of the distribution of DCT residual energy to find candidate dissolve shots, and then it uses motion vectors and direct current (DC) component's histograms to identify those candidates. Experimental results show the proposed method is much fast, robust and of high accuracy.2. Propose a qualitative analysis method of dominant motion for compressed video stream: It makes use of the directional distribution of motion vectors and the relationship between any two motion vectors in MPEG-1/2 stream. It is possible to obtain reliable estimation of dominant motion by directly processing motion vectors obtained from the compressed stream.3.Implement a video scalable coding method in frequency domain: Thetraditional video coding method is fragile to transmission error over networks. Scalable video coding is an effective method to cope with this problem. Based on MPEG-2 data partition, we implement a video scalable coding method in frequency domain, which decreases the redundancy resulting from data partition. Furthermore, the thesis gives a method for data synchronization between base layer and enhancement layer.4. A novel error concealment algorithm based on shot boundary detection is proposed: There are two conventional approaches to error concealment, i.e. temporal error concealment and spatial error concealment. Due to low computational complexity and good concealed image quality, temporal error concealment is the preferred way of error concealment. However, temporal error concealment will be apparently ineffective at shot boundary. To address this problem, we propose a novel error concealment algorithm, which adopts real time shot segmentation to detect whether or not current damaged frame is a shot boundary, utilizing information acquired during decoding process. In the case of shot boundary, the damaged frame is concealed by a spatial error concealment method; otherwise a temporal method is performed.5. Proposed an efficient rate control method for MPEG-2 to H.264/AVC transcoding: H.264/AVC standard recommends a rate control method, which is based on a quadratic rate distortion model, i.e. VM model of MPEG-4. The Quantization parameters for I- and B-frame are simply interpolated from known quantization parameter for P-frame. Taking advantage of the character of quantization parameter in H.264/AVC, we proposed a rate control method for MPEG2-to-H.264/AVC transcoding. Quantization for P-frame is obtained from quadratic model, while quantization parameters for I- and B-frame are computed based on TMN5 model. Objective experiments and subjective evaluation indicate that the proposed method is superior to rate control method recommended by H.264/AVC in both smoothness of bit rate and PSNR of image.6. We design and implement an IPTV system which provides demanding service and broadcast service. This IPTV is composed of two sub-systems: Content Based TV Browse and Retrieval (CBTVBR) sub-system and Internet DTV/HDTV sub-system. CBTVBR provides some functions such as video content analysis, digital TV program retrieval and browse etc, while Internet DTV/HDTV sub-system provides broadcast service over multicast networks. Those two sub-systems integrate all key techniques...