| Video coding is one of the kernel technologies of broadband network stream media, wireless broadband multimedia communication and internet new multimedia applications. Information on the network due to the increasing demand for video transmission, the compression video data not only must face the memory, but also must face each kind of question which in the network transmission possibly meets. Heterogeneous network, bandwidth fluctuations, the reception and transmission of broadcast equipment reliability and diversity gave the traditional fixed rate based on the optimization of video coding technology as a challenge. In the network environment, the demand for video coding is simply no longer pursuing a fixed bit-rate coding efficiency optimization. Instead, under the premise of guaranteeing high compression efficiency, compressed video stream can fulfill the requirements of a certain area within the bandwidth for better video quality. Namely has the code rate scalability to adapt the network band width change, varied demand for services and other factors.Scalable Video Coding recent decades has been a hotspot. Scalable Video Coding could produce the embedded code stream. The smallest intercept bit stream (base layer bit stream) rate is usually lower; it is an independent receiver to decode. When the user receives more bit streams (enhancement layer bit stream), they are decoded together with the basic level bit stream, and the user may obtain the better decoding video quality. Scalable coding in the bit stream can be interrupted in the media server transmitting end and in the transmission process, adapts the network environment band width change.The wavelet transform is developed in the past 10 years as a signal analysis method. Along with the continuous deepening of wavelet image coding technology,wavelet has been found a good time-frequency analysis and the natural characteristics of a multi-resolution structure, and coincide with the characteristics of human vision. This makes the video coding technology based on wavelet compression with a good performance and a high degree of scalability, not only can avoid traditional the video code algorithm using the DCT block method inherently to bring "the block effect", the fault-tolerant performance especially is also splendid. This enables the wavelet coding technology to have the possibility to become one kind of good expandable video frequency code solution.If to some kind of code method production bit stream, a single code stream decoder can be extracted from some of the contents to achieve a certain rate spatial resolution and frame rate required to restore the quality of video coding algorithm, this code algorithm is called the fully scalable video coding algorithm. As a result of the wavelet transformation multi-resolution characteristic, and it's specially features for embedded coding, causes it has the natural superiority in the resolution and the code rate scalable characteristic aspect. Compares the tradition motion compensation prediction mix video coding framework, in the temporal domain with a one dimensional wavelet transform, completed in two-dimensional spatial structure of the change of wavelet decomposition can provide SNR (quality) temporal and spatial three kinds of scalable properties. Sub-band coding can be removed by low-resolution sub-band dependence on the high-resolution sub-band. As for the one-dimensional transform in the time-domain, the closed loop prediction is changed open-loop motion compensation, therefore under some quality level, avoided arranging the end of the codec to cause the drifting phenomenon to the reference object change, ultimately reduce the occurrence of the decline in the quality of the reconstructed video. In addition, temporal filter relatively during simple frame, the multi-frames prediction the way can use the video well in the time direction redundancy, thus enhances the video coding performance. Based on the above analysis, This is the direction in terms of time and space to make the transformation filtering method is particularly suited to the selection of fully scalable video bit stream. Because of this, this paper has developed based on the wavelet fully scalable video coding method research.In the beginning of this paper, the first chapter mainly introduced the Scalable Video Coding research background and significance. In the second chapter, the scalable video coding technology and the lifting scheme Wavelet theory is briefly introduced. Given domestic and foreign Scalable Video Coding research status and development direction, in detail introduced the scalable video coding based on wavelet scheme, and gives some typical coding framework.Based on the multi-resolution wavelet transform characteristics and special features for embedded coding, this paper uses the motion compensation 3D wavelet transformation realization completely scalable video coding. In the third chapter gives a based on the lifting wavelet motion compensation temporal filter technology fully scalable video coding system framework, and bit stream and the organizational structure is designed to enable them to take elements from arbitrary bit stream decoder to meet the requirements of the scalable parameters. The motion compensation temporal filter technology is one of most important parts in the wavelet video coding system, it can effectively remove the redundant information in temporal domain, therefore has a very significant impact on the performance of the code. This paper uses based on the hierarchical variable block size motion estimation method realizes the movement to estimate. The central idea is that estimates the high resolution picture movement vector through the low resolution picture movement vector. This is because the low-resolution image motion vector estimation can be used small regional scope of the search, and the high resolution image movement estimated may use the estimate movement vector of the low resolution picture to take the search window the center, therefore by this center, in a small region-wide search, the search motion estimation algorithm thus greatly reduce the amount of computation. After the movement vector is obtained, along the path of motion estimation 5/3 lifting wavelet is used to carry on the temporal filter. 5/3 lifting wavelet transformation not only keeps the tradition wavelet transformation superiority, effectively removes the time redundant information, but also can realize the integer to the integer wavelet transformation, achieve rapid, low computational complexity can be improved wavelet video coding performance. To improve the coding efficiency and the elimination of the effects of the introduction of coding, aiming at the pixel of the different types of connecting this paper adopt the different processing method, and according to the motion compensation effect auto-adapted choice temporal domain decomposition progression .In the 3D sub-band coding, this paper used the embedded variable zero-block coding algorithm with a quad-tree structure expression. The algorithm is based on the classical two-stage bit-plane coding algorithm, among the coding method, a quad-tree decomposition skill is used to express the variable size zero block in order to improve the performance of the significant map coding.In the fourth chapter the software simulation realization has been carried on to this coding framework. Through comparison with other video coding and spatial, temporal, SNR scalability test results this paper presents the algorithm has proved not only completely scalability, but also the characteristics of high-performance compression. Finally in the fifth chapter, the summary to this paper and future work prospects are made. |
PDF Full Text Request