| In recent years, with the development of social informatization and networkbandwidth, digital multimedia technology has been greatly developed and foundwidely applications, which is gradually influencing and changing the world. With trueand natural reproduction of the scenarios for the users, multi-view video has attractedmuch attention of academia and industry, and becomes one of the hotspots in the fieldof video research. The multi-view video contains a huge amount of data, which is agreat challenge to data storage and transmission, therefore it needs efficientlycompressed. Multi-view video coding eliminates redundant information by usingsome characteristics of the multiview video images, reducing the mass of data. Thekey technologies involve the prediction structure, motion estimation and disparityestimation, mode selection and three-dimensional residual image coding, occlusiondetection, and so on. In this paper, two key techniques of the multi-view video coding,prediction structure and fast mode decision, are studied, and two innovativealgorithms are proposed. The main contents and results are as follows:1．An adaptive prediction structure encoding scheme is proposed in this paper.Four different prediction structures have been designed based on the relationshipbetween the temporal correlation and the inter-view correlation of multiview videosequences.The optimal prediction structure is selected according to the characteristicsof the2D GOP. By using flexible prediction structure, the number of the referenceframe is reduced, and the encoding complexity is lowered. Consequently, theencoding efficiency is significantly improved.2. A novel mode decision method based on mode classification is proposed. Themulti-view video coding improves the compression efficiency by selecting the bestcoding mode with the rate-distortion optimization strategy. However, thecomputational complexity of the exhaustive mode searching is very high, especiallyin the sub-macroblock mode searching. To solve this problem, a mode classificationbased early termination fast mode decision method is proposed to quicken theencoding speed on the basis of the statistical analysis of the macroblock modeselection characteristics in the multi-view video. |
PDF Full Text Request