Three-dimensional broadcast television (3D TV) is the next generation multimedia application. The European ATTEST project proposed using monoscopic video plus depth map information instead of stereoscopic video to produce 3D video.
Due to the local smoothness characteristic of most real-world object surfaces, the per-pixel depth information can be efficiently compressed instead of using an additional color video channel. This characteristic makes it possible to introduce a 3D TV service based on the proposed concept with only a very small transmission overhead (below 10-20% of the basic color video bitrate) compared to conventional 2D digital TV system.
However, the complexity and hardware requirement are nearly two times higher than coding 2D video. In 3D videos, some relationships such as motion vectors and object boundaries between texture video and depth map sequence can be used to reduce the complexity. By using sharing-of-motion-information method and analyzing these relationships, we propose a proper hardware architecture for coding depth map sequence with low complexity and hardware cost.
In this thesis, we have proposed a H.264 based depth map sequence coding by using predictive motion estimation algorithm with sharing motion information. We also propose a hardware design of depth map sequence coding by using our algorithm.
In the software algorithm, we propose predictive motion estimation algorithm to modify the texture motion information and mode reduction to reduce the complexity. In our algorithm, it doesn’t need any additional effort on the decoder, and the image quality can also be approximate to original H.264 at low bit-rate. At high bit-rate, our algorithm outperforms other sharing of motion information algorithm.
In the hardware design, due to complexity reduction we use 8x8 PE2D to implement predictive motion estimation. We need additional SRAM and Sobel filter to implement, but 3/4 PE2D is saved. The overall depth map sequence encoder can be smaller than original H.264 encoder. We also solve the bubble cycles problem at 8x8 PE2D processing time.
The prototype design is implemented using Xilinx multimedia board and the design limitation is 100MHz in the worst case, which is higher than the required frequency, 68.8MHz, in 720x480 size frame.

立體廣播電視系統被視為下一代的多媒體應用。歐洲的ATTEST專案提出使用單一視訊加上深度資訊進而產生3D視訊來取代傳統的傳輸雙視訊的系統。
由於現實世界多半物體的面都是平滑的，不至於會有太多的深度變化使深度資訊比起一般的視訊更能夠有效率的被編碼。比起傳輸一般視訊，傳輸單一視訊加上深度資訊產生立體視覺只需要多一小部分的傳輸負擔，使立體視訊系統更能普遍的應用於現實生活。
然而，由於硬體需求及編碼複雜度仍高於編傳統的視訊。所以如何降低複雜度以及硬體需求是需要被考慮的。在深度資訊中仍然有一些動作資訊以及物體邊緣的特性和紋理視訊有關，所以便有利用共用動作資訊的方法被提出。透過這個方法以及去分析深度資訊的特性，我們提出了一個適合做為硬體使用的演算法及架構，用來降低複雜度以及硬體需求。
在這篇論文中，在基於H.264的架構下，我們提出了利用紋理的動作資訊來預測深度資訊動作資訊方法，以及一個根據這個方法而建構出的硬體架構。
在軟體演算法上我們除了利用紋理的動作資訊來預測深度資訊動作資訊來降低複雜度外，我們同時計算比較少的模式來更近一步降低複雜度。在我們的演算法中，不需要在解碼端加入額外的硬體便可以達成。在低位元率需求時，我們的編碼效率可以逼近於原本的H.264編碼。而在高位元率時，我們的編碼效率可以優於其他使用共用紋理動作資訊的方法。
在硬體設計中，由於複雜度的降低，我們可以只用8x8PE2D架構，雖然我們需要額外的SRAM和Sobel濾波器。比起原本的16x16計算架構上，我們只需要原本的1/4的16x16PE2D架構面積，整體的硬體需求還是比原本的來的低。在設計同時我們也解決的在8x8計算時會產生的垃圾週期。
我們使用Xilinx多媒體版來實踐設計的原型而設計的最大可操作頻率為100MHz，大於720x480大小的序列所需的68.82MHz。

[1] C. Fehn., “Depth-Image-Based Rendering (DIBR), Compression and Transmission for a New Approach on 3D-TV”, In Proc. of SPIE Stereoscopic Displays and Applications XV, San Jose, CA, USA, January 2004.
[2] C. Fehn, N. Atzpadin, M. M□ller, O. Schreer, A. Smolic, R. Tanger, and P. Kauff, “An Advanced 3DTV Concept Providing Interoperability and Scalability For a Wide Range of Multi-Baseline Geometries”, In Proceedings of International Conference on Image Processing, pages 2961-2964, Atlanta, GA, USA, October 2006
[3] N.A. Dodgson, “Auto-stereoscopic 3D displays”, Computer Volume 338, Issue 8, Aug 2005.
[4] S. Grewatsch and E. Muller, “Sharing of motion vectors in 3D video coding”, International Conference on Image Processing, Oct. 2004.
[5] Han Oh, and Yo-Sung Ho, “H.264-based Depth Map Sequence Coding Using Motion Information of Texture Video” PSIVT 2006 898-907.
[6] M. T. Pourazad, P. Nasiopoulos, and R. K. Ward, “An H.264-based video encoding scheme for 3D TV”, EURASIP European Signal Processing Conference -EUSIPCO (Florence, Italy), September 2006.
[7] C. Fehn, K. Hopf and Q. Quante, “Key Technologies for an Advanced 3D-TV System”, Proceedings of SPIE Three-Dimensional TV, Video and Display III, pages 66-80, Philadephia, PA, USA, October 2004. (Invited Paper)
[8] "JM12.4 Test Model CODEC," ISO/IEC MPEG and ITU-T VCEG Joint Video Team, http://iphome.hhi.de/suehring/tml/
[9] C. Fehn, K.Schuur, I. Feldmann, P. Kauff, and A. Smolix, “Distribution of ATTEST test sequence for EE4 in MPEG 3DAV”, ISO/IEC JTC1/SC27/WG11/M9219, Dec. 2002.
[10] T. Wiegand, G. J. Sullivan, G. Bjontegaard, and A. Luthra, “Overview of the H.264/AVC video coding standard”, IEEE Trans. Circuits Syst. Video Technol., vol. 13, pp. 560-576, July 2003
[11] T. Komarek and P. Pirsch, “Array Architecture for Block Matching Algorithm”, Circuits and Systems, IEEE transactions on Volume 36, Issue 10, Oct 1989.
[12] XILINX, “Virtex-II Platform FPGA User Guide”, UG002(v1.4) Nov. 1, 2002.
[13] XILINX, “MicroBlaze and Multimedia Development Board User Guide”, UG020(v1.0) Aug. 29, 2002.