1.pdf2.pdf3.pdf3.pdf66.pdf4.pdf4:2:2 Profile @Main Level (Pages: 192-196). '. *- b > --4 * MPEG-2 4:2:2 Profil@,Main Level Prof~sional studio/p\ost-production environment s generation, chroma keying using compressed video s, 4:2::0 format). Higher bit 2:O and 4:2:2 formats. Good environments. Vx robust in - sparent or near transpar6nt in chroma Level Compatibility. No scalability ess which uses a switching signal to blend 4 btra dc coefficient precision: up to 1 1 bits S-e (Intrastuciiol BPl%~~sep - et C @P Long GOP structure (Interstudio) + - '- E!BP .......(N= 15 & M=3 or N=12 & M=3) luminance and chrominance quantization tables -- - : 2' Upper bounds for spatiaVtempora1 resolutions 720 pelslline, 608 linedfiame 4 ; at 50 HZ and 720 pels/line 512 lines/fiame at 60 HZ - Max VBV Buffer size 9437 1 84 bits FrameEield DCT ,, FrameEield MEJMC , L - ' . * - n.- -5 'w. e a m 8 h 2 i: P, 3References Video Sub-group and Requirements Sub-group, "4:2:2 MPEG-2 profile proposal: Programme of work," in Int. Organization for Standardization, Singapore, Nov. 1 994, ISO/IEC JTC llSC291WG11 N0858. ISO/IEC 1381 8-2 Amendment 2, Generic Coding of Moving Pictures and Associated Audio, International Organization for Standardization, Jan. 1996. C. Horne, et al, "Multigeneration experiments using 4:2:2 profile," in Int. Organization for Standardization, Lausanne, Switzerland, Mar. 1995, ISOIEC JTCllSC29IWGll MPEG951032 T. Naveen, et al, "On improving the efficiency of VLC's for 4:2:2 MPEG-2 profile," in Int. Organization for Standardization, Lausanne, Switzerland, Mar. 1995, ISOIEC JTC l/SC29/WG11 MPEG951063. T. Naveen et al, SPIE, Photonics East, Vol. CR-60, pp. 288-308, Philadelphia, PA, Oct . 1 995. C. Horne et al, "Study of the characteristics of the MPEG-2 4:2:2 profile - Application of MPEG-2 in studio environment " v 1. 6, pp. 251-272,June1996 TAMS. CS- rg.Multi-View . . Profile (Pages: 197-213)Document N1366 Sept. 1996. Also ITU-T Rec. H.262 (1 995)IAMD3 (1 996E) AMD: Amendment Multi-View Profile (MVP) Supports stereoscopic pictures for a wide range of picture quality & resolution Depth Perception Base layer-Left View Enhancement layer-Right View Base layer: Monoscopic coding with the same tools as Main Profile Enhancement layer: coded using temporal scalability tools and a hybrid prediction of motions and disparity. PROFILE LEVEL Main Low ISO/IEC 13818-2 Amendment 3 (VIDEO) MPEG2 X. Chen and A. Luthra "MPEG-2 multiview Profile and its application in 3D TV" SPIE/IS & T Elect~onic Imaging: Science and Technology vol. 3021, San Jose, CA, Feb. 97. 2. Con.. 302 1 A, "Stereoscopic displays and Applications VIII," SPIEIIS & T, San Jose, CA, Feb. 97. - : 3. Session on c'Stereoscopic Processing & Analysis" SPIE/IS & T, vol. 3024, San Jose, CA, Feb. 97. ' .Ti . 4. A. Puri, "Video Coding Using the MPEG-2 Compression Standard," Proceedings of SPIE Visual Communications and Image Processing, pp. 1701-1713, Boston, MA, Nov. 1993. 5. A. mui, R. V. Kollarits and B. G. Haskell, "Stereoscopic video compression using temporal scalability," SPIEIVCIP, vol. 2501, pp. . . 745-756, Taiwan, Taipei, May 1995. -La'3D Video constructed from two views of 2D video from different perspectives. Autostereoscopic system. Two views hsed in the brain to provide perception of depth. length Left Camera Right Camera 1 a! Figure 1. Stmoscop~c imaging and Display geometries Scene Right Canera, Viewer Stereoscopic Stereoscopic '3 Encoder Decoder + Left Camera Figure 4 A Stenoscopic video system - - -1 . ' -" ' .. .,. .I..;,- -. ;..%%, ,. . ?C'. .- 8GA , ,-* ..nL'- , : .-gjg .:,+$? 3-t.. .*- "; , t': I ', ;, , . .. . ?-8. r&. ? . -. -&,-- .a . . ~ - *-- -)- -----I h i. Dis~aritv Com~ensated Prediction: Configuration 1 I '~em~ord coincidence between the left view and right view. Enhancement Layer Base Left fl fl fl (p-I Layer PyY t - Fig. 5 prediction configuration 1 with M=3 coding of left-view, right-view fiame picture coded using dispaity predictions with respect to left-view. Dis~aritv and motion compensated prediction configuration 2. Enhancement Right Layer View AWA WAWAW I I I I Left 0 (J 8--* View . Fig.6 Prediction configuration 2 with M=3 coding of left-view, right-view fiame picture coded using disparity prediction with respect to left-view and motion prediciton with respect to itself.Tem~oral Scalability in bi-directional prediction configuration 1: Enhffncement layer B pictures only (forward, backward & bi-directional MCP) Right View Left View Fig. 2 Prediction configuration 1 with M=3 fiame-picture coding of baseIayer 5 using frame-picture both predictions fiom base-layer. Configuration 2: Enhancement layer: Mainly B pictures (first picture is a P picture) (forward, backward & bidirectonal MCP) -- kft Base View layer Fig, 3 Prediction configuration 2 with M=3 he-picture coding of base-Iayer, &cement-layer, enhancement-layer using fiame pictures with one prediction fkm enhancement-layer end one fiom base-layer.Compensated DCT Right View Encoder Mux Motion Compensated DCT Left View Encoder SYS Demux Motion & Disparity 1 Compensated DCT Right View Decoder Motion Compensated DCT Left View Decoder Fig. 4(a) A generalized stereoscopic video codec derived from temporal sealability codec structure of Fig.1. Motion and Dispanity Compensated DCT Encoder ....................... Compensated W I DC T Encoder 1 Fig 4(b) Cocks of I?ig.4(a) revised to show some Motion and Disparity Compensated DCT Decoder Motion DC T Encoder details of combined disparity and motion compensated prediction. Disparity estimation and compensation based on (16x16) Y block.Sys. Demux Enhancement Decoder Out Enhancement Layer SYS Mux. Enhancement Encoder Temporal Demux + Lower I1 I b I Decoder I A Lower Layer In A Two Layer Codec Structure for temporal Scalability Temporal Demux Lower Encoder Left View L R Right View - Disparity Stereo disparity fusion methods b Time Unit I Motion Motion 1 Time Unit I + 1 1 b v v DisparityDisparity prediction by use of "most recent decoded enhancement picture" and 1 or "most recent lower layer picture order". Time P Most recent decoded enhancement ', I B . Most recent Iower layer picture in display order 4-7 Iower layer Enhancement Layer Predict Modes for P and B-PicturesUpper bounds for bit rates (Mbitls) The base layer of the MVP is assigned to a left and the enhancement layer is assigned to a right view. A monoscopic coding with the same tools as Main Profile (MP) is applied to the base
View Full Document
Unlocking...