Techniques for low bit-rate video coding

Abstract

Digital video compression techniques play an important role in the multimedia information era. Hence, the objectives of this research work are to devise and develop efficient methods for video coding. In this thesis, different approaches for video coding such as block-based motion compensation coding scheme, and foreground/background video coding scheme will be described. Furthermore, several successful video coding standards such as H.261/H.263, and H.263+ will be reviewed in this thesis. Due to the nature of block-based coding, blocky artifacts may appear on the compressed video. We will also review some advanced video coding algorithms that solve this type of problems. As for the foreground/background video coding scheme, the first step is to segment out the targeted object in a video sequence. Since the major component for many applications, such as videophone and video conferencing, is the human face, we will also provide an overview of the techniques. In our research, we proposed an efficient algorithm for human face detection and facial feature extraction. The genetic algorithm and the eigenface technique are employed for detecting the face and facial feature regions in our approach. The genetic algorithm is applied to search for possible face regions in an image, while the eigenface technique is used to determine the fitness of the regions. As the genetic algorithm is computationally intensive, the searching space is reduced and limited to the eye regions in order to substantially reduce the computational time. Possible face candidates are then further verified by measuring their symmetries and determining the existence of the different facial features. Furthermore, in order to improve the level of detection reliability in our approach, the lighting effect and orientation of the faces are considered and their related problems solved. A very low bit-rate video coding algorithm by focusing on moving region has also been proposed. Eight patterns are pre-defined to approximate the moving regions in a macroblock. The patterns are then used for motion estimation and compensation to reduce the prediction errors. Furthermore, in order to improve the compression performance, the residual errors of a macroblock are rearranged into a block with no significant increase of high order DCT coefficients. As a result, both the prediction efficiency and the compression efficiency are improved. Finally, a foreground/background video coding system is developed for application of the videophone. We have also incorporated information about the facial features in our moving region based coding algorithm in order to provide a better picture quality perceptually.