現今視訊會議、廣告輪播以及現場演講都需要播放影像內容，透過大型電子顯示設備，如電視、投影機等設備。然而隨著時代的演變、科技的進步，人們希望這些電子顯示設備擁有更大的尺寸、更高的解析度。但是現在這些大尺寸、電子顯示設備動輒數十萬，像是鴻海一百二十吋電視亦或者是索尼的一百吋電視。要進行大型電子顯示設備的影像播放，最經濟的方式就是使用電視牆，也就是拼接多台電子螢幕顯示器。在多台拼接螢幕顯示器上，由於影像資料的檔案容量以及個人電腦硬體裝置的規格限制上，如何將視覺內容快速地讀取並且即時地顯示至對應的顯示設備上，是這個研究的主要目標。本篇論文提供了一種新穎的影像拼接系統，用於快速讀取資料並即時解壓縮影像且同步視覺內容渲染至不同顯示設備上。該系統在解壓縮階段使用多圖形處理器（Multi-GPU）加速輸入影像的解壓縮效能。接著於渲染階段，拼接數個已處理好的高畫質（SuperHD/4K）色彩資訊，並且無縫同步地顯示在電子顯示設備上，呈現超高解析度（8K/16K）的色彩資訊。本篇提出了兩大重點。首先，利用多圖形處理器（Multi-GPU）針對高畫質（SuperHD/4K）以上的S3 紋理壓縮（DXT）或圖像壓縮（JPEG）進行解壓縮動作。另一部份，利用圖形處理器（GPU）的記憶體，規劃同步佇列緩衝器（Synchronized Queue Buffer），儲存已解壓縮之色彩資訊，取出影像拼接成更高的解析度，並同步顯示。在本篇論文最後的研究結果中，證明了拼接數張高畫質（SuperHD/4K）解析度影像可以於一般電子顯示設備上即時播放，且幀速率（Frames Per Second，縮寫為FPS）可以到達30-60 以上。本篇提出的新型影像拼接系統為未來的超高畫質（8K/16K）影像提供了令人滿意的觀看性能。

This paper presents a novel video stitching system for stitching multiple 4K decompressed images and displaying the seamless videos with four/eight times the resolution of the SuperHD/4K video. The key insight of the proposed system with two stages is twofold. First, for decoding the 4K videos with the DXT and JPEG compression efficiently, multiple GPUs are utilized to ameliorate the performance of the video decoding. Then, the strategy of the synchronized queue buffer are implemented to store the decompressed 4K images for stitching synchronized video frames with four/eight times the SuperHD/4K resolution. In addition, to prevent the performance decrement of transmission between the CPU and GPU, the original input video with the DXT and JPEG compression is uploaded to the GPU memory in advance. The experimental results demonstrate the stitched seamless videos with four/eight times the 4K resolution can be displayed on the physical electronic display devices in real time, and the frame rate of the resultant videos achieves 30-60 frames per second (FPS). Consequently, the proposed novel video stitching system provide satisfied performance of video watch for the future generation of 8K/16K video processing systems.

[1] Petr Holub, Martin Srom, Martin Pulec, Jiri Matela, and Martin Jirman. GPUaccelerated DXT and JPEG compression schemes for low-latency network transmissions of HD, 2K, and 4K video. Future Generation Computer Systems, 29(8):1991 – 2006, 2013. Including Special sections: Advanced Cloud Monitoring Systems & The fourth IEEE International Conference on e-Science 2011 – e-Science Applications and Tools & Cluster, Grid, and Cloud Computing.
[2] P. Brown. S3 texture compression. version 1.5, NVIDIA Corporation, 2009.
[3] Ngai-Man Cheung, Xiaopeng Fan, O.C. Au, and Man-Cheung Kung. Video coding
on multicore graphics processors. Signal Processing Magazine, IEEE, 27(2):79–89, March 2010.
[4] H. Malvar and G. Sullivan. YCoCg-R: a color space with RGB reversibility and low dynamic range, 2003.
[5] J. van Waveren and I. Castao. Real-time YCoCg-DXT compression. Tech. Rep., id Software Inc., NVidia, 2007.
[6] P. Patel, J. Wong, M. Tatikonda, J. Marczewski, JPEG compression algorithm using CUDA, Tech. Rep., Department of Computer Engineering University of Toronto, 2009.
[7] A. Obukhov, A. Kharlamov, Discrete cosine transform for 8x8 blocks with CUDA, Tech. Rep., NVidia, October 2008.
[8] S. Tokdemir, DCT implementation on GPU, Master’s Thesis, Georgia State University, 2006.
[9] Z. Hong, K.I. Iourcha, and K.S. Nayak. Fixed-rate block-based image compression with inferred pixel values, August 10 2004. US Patent 6,775,417.
[10] J. Matela, M. Šrom, P. Holub, Low GPU occupancy approach to fast arithmetic coding in JPEG2000, in: MEMICS 2011, in: Lecture Notes in Computer Science, vol.
7119, Springer, Berlin, Heidelberg, 2012, pp. 136–145.
[11] ITU-T recommendation T.81, published by International Telecommunication Union, CCITT, Telegraph and Telephone Consultative Committee, Sepetember 1992.
[12] E.J. Delp and O.R. Mitchell. Image compression using block truncation coding. Communications, IEEE Transactions on, 27(9):1335–1342, Sep 1979.
[13] ITU-T recommendation T.81, 1992.