Free Viewpoint Generation from H.264/AVC Decoding Procedure

• Main Authors: Hui-Leng Fu, 傅惠稜
• Other Authors: Jar-Ferr Yang
• Format: Others
• Language: en_US
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/18509608948764543708

碩士 === 國立成功大學 === 電腦與通信工程研究所 === 96 === Researches in both computer vision and graphics have made important steps toward generating free view video sequences. Free viewpoint video offers the viewers an opportunity to interactively change its viewpoint in the scene. Existing methods for generating new views can be broken down into two basic groups: Model-based rendering (MBR) and Image-based rendering (IBR). These two kinds of method both need a lot of computation. In this thesis, we develop an image-based free viewpoint sequence generation method. It is based on the post processing of H.264/AVC decoding. Through the information of motion vector after decoding, the computation complexity can be reduced. This scheme contains three parts: 1) disparity map estimation, 2) virtual view interpolation, and 3) hole filling. Experimental results show that this approach can generate virtual view between two real views that have horizontal disparity with different weighting factor through the whole sequence.