基于光场渲染多视点视频编解码方法研究.pdfVIP

4 08 V o l. 40, No. 8 2 0 1 0 8 JOURNAL OF UNIVERSIT Y OF SCIENCE AND T ECHNOLOGY OF CHI NA A ug. 2 0 1 0 : 0253277 8( 2010) 0808 1206 程 龙, 郭 立, 陈晓琳, 袁红星 ( , 230027) : 首先提出一种面向移动设备的多视视频纹理编码方法, 编码端利用重建的3D 网格信息 4D 光场重采样, 得到顶点的光场矩阵; 然后提出一种基于奇异值分解( SV D) 的自适应分解级数选 取方法, 保证真实感的同时尽可能减少光场数据量. 为了配合解码端的图形硬件加速, 采用了 S3T C 纹理压缩标准 分解后的矩阵进 一步压缩, 减少编码端数据量以及解码端纹理渲染的带宽 需求. 实验结果表明, 在保证较好的渲染效果下, 提高了压缩比, 并且满足实时渲染要求. : 多视点视频; 光场渲染; S3T C; 自适应奇异值分解 : T P391. 4 1 : A do : 10. 3969/ j. ssn. 0253277 8. 2010. 08. 00 The research of a multiview video encoding and decoding method based on light field modeling CH EN G Lo ng , GU O L , CH EN X ao l ng , Y U A N H o ng x ng ( D ep t. of E lec tr oni c Sc ienc e and Te chno logy , Univ ers i ty of S ci ence and T ech nol ogy of Chi na, H ef ei 230027, Chi na) Abstract: A met ho d for v deo t ext ure encod ng and deco d ng o n a m ob le dev ce w as proposed . On t he encoder s de, t he 4 D l ght f eld w as resampled based on t he reconst ruct ed 3D model and then the vert ex l ght f eld mat r x w as acqu r ed . A n opt m zat on was to adjust t he deco mpos t on lev el selfadapt vely based on SV D t o decrease t he data quant t y w h le guarant ee ng the 3d ment onal ty n l ght f eld matr x f act or zat o n. In order t o ut l ze graph cs hardw are accelerat on , S3T C t ex t ur e compress o n st andard w as u sed t o furt her decr ease dat a quant t y at t he encoder end and bandw dt h demand at t he decoder end. T he exper m ent result s show t hat t he proposed met hod can m pro ve t he compress o n rate w h le guar ant ee ng bet t er render ng eff ect , and meet t he demands of realt me r ender ng.