透過您的圖書館登入
IP:3.145.8.42
  • 學位論文

航照影像校準之平行計算

Parallel Computing of Aerial Image Registration on GPU

指導教授 : 謝東儒

摘要


影像校準(Image Registration) ,是電腦圖學的一項應用。目的是把兩張(或以上)有場景重疊影像,依照特徵合成為一張圖檔的程序。但隨著科技的進步,影像解析度的提高造成處理時間大增,舉例來說計算1024 ×1024的圖檔將比512×512的圖檔多出4倍的計算量,如果用一般的處理器計算,效能已經不足,因此本論文提出一個方法,改良過去單線程(Single thread)的影像校準程序程序,並利用顯示晶片強大的計算資源與記憶體頻寬進行平行化,以提升計算效率,並比較其效能的差異。因為所得到的圖檔都為高解析度,因此在影像校準處理完後會將結果透過平行顯示技術在4×4的顯示器陣列上已達成高解析度顯示成果。

關鍵字

GPGPU 影像校準 GLSL CUDA 影像處理 平行顯示

並列摘要


Image registration is an application of Computer Graphics. This process overlaps two images that have same features. The input images may been taken from different angles or different viewing directions. With the advance of the hardware technology, the resolution and size of image data becomes larger. It would require long computation time using CPU. We have developed image registration techniques based on programmable graphics processor to increase performance. Finally, we successfully visualize the result images on large LCD display walls.

並列關鍵字

GPGPU Image Registration GLSL CUDA Image Process Parallel Rendering

參考文獻


[3] Timothy J. Purcell, Ph.D. Thesis, RAY TRACING ON A STREAM PROCESSOR, Stanford University, March 2004.
[6] Mark Harris, “Mapping computational concepts to GPUs, ” ACM SIGGRAPH, 2005
[9] Schatz, M.C., Trapnell, C., Delcher, A.L., Varshney, A., “High-throughput sequence alignment using Graphics Processing Units., ” BMC Bioinformatics, 2007.
[10] Svetlin A. Manavski and Giorgio Valle, “CUDA Compatible GPU Cards as Efficient Hardware Accelerators for Smith-Waterman Sequence Alignment, “BMC Bioinformatics, 2008.
[16] John D. Owens, David Luebke, Naga Govindaraju, Mark Harris, Jens Krüger, Aaron E. Lefohn and Timothy J. Purcell, 2005. A Survey of General-Purpose Computation on Graphics Hardware. Eurographics 2005, State of Art Report.

延伸閱讀