針對工業製程瑕疵檢測，傳統的二維自動光學檢測（Automatic Optical Inspection）方式，通常僅能檢測物件表面缺陷。故對於內部缺陷檢測可利用有穿透能力的X射線得到可能的解決方案。
藉由電腦斷層掃瞄（Computer Tomography，CT）進行物件三維結構的影像重建，利用得到之立體影像來判斷內部缺陷。
對於把二維的投影影像重建成三維影像常見的演算法有兩類，一類為濾波逆投影法（Filtered Back-Projection，FBP)，另一類為代數重建法（Algebraic Reconstruction Technique，ART）。在投影數較少時，利用代數重建法（ART）所得影像要比利用濾波逆投影法（FBP）好的多。但是，因代數重建法（ART）影像重建的速度太慢，在運算速度上無法達到使用者的需求。
透過多核心處理器的普及下，使得繪圖處理器（Graphic Processing Unit，GPU）也邁入多核心架構。讓新一代的繪圖處理器（GPU）結合中央處理器（Central Processing Unit，CPU），強調高平行度以及優越的運算能力。因此，本研究在維持影像品質的考量下，達成提高加速運算之目的，提出平行處理方式實作於繪圖處理器(GPU) 與簡化權重運算。

Traditional, two-dimensional automatic optical inspection methods can be only used for detecting surface defects and cannot be used for detection of internal defects. Therefore, the X-ray Computer Tomography (CT) method has been used for re-construction of three-dimensional images of the inspected objects, and also the detection of the internal defects. However, the image re-construction process is time-consuming. There are two common ways for image reconstruction: one is Filtered Back-Projection (FBP), the other is Algebraic Reconstruction Technique (ART). When the number of projection is little, using ART can get better image quality. However, ART takes longer time for image reconstruction. This paper focuses on speeding the computation by using hardware with parallel processing and simplified weight. We implemented the parallel computing of iterative reconstruction Method in General-Purpose computing on Graphics Processing Units (GPGPU).
Keywords：X-Ray Computer Tomography, ART, GPU

[1] AMD Functional Data Sheet, 754 Pin Package, Advanced Micro Devices, New York, 2004.
[2] T. D. Moore, “Three-Dimensional X-Ray Laminography as A Tool for Detection and Characterization of BGA Package Defects,” IEEE, Transactions on Components and Packaging Technologies, Vol. 25, No. 2, pp. 224-229, June 2002.
[3] Avinash C. Kak and M. Slaney, ”Principles of Computerized Tomographic Imaging,” IEEE Press, 1988.
[4] Gabor T. Herman, “Image Reconstruction From Projections：The Fundamental of Computed Tomography,” Academic Press, 1980.
[5] A.H.Andersen and A.C.Kak, "Simultaneous algebraic reconstruction technique （SART)：A superior implementation of the ART algorithm," Ultrason. Imaging, vol. 6, pp. 81-94, 1984.
[6] F. Xu, “Rapid Tomosynthesis with Exact Ray-driven Projector using Graphics Processing Units,” 2nd Workshop on high Performance Image Reconstruction(HPIR) , 2009
[7] 林源益，「應用同步代數重建法於BGA檢測之研究」，國立清華大學動機所，2008年6月。
[8] 梁志彰，「應用FDK重建法於BGA檢測之研究」，國立清華大學動機所，2008年6月。
[9] S. Krimmel, J. Stephan, and J. Baumann, “3D Computed Tomography Using A Microfocus X-ray Source: Analysis of Artifact Formation in The Reconstructed Images Using Simulated as well as Experimental Projection Data,” Nuclear Instruments and Methods in Physics Research, Section A-542, pp. 399-407, February 2005.
[10] R. Ning, X. Wang, and D. L. Conover,“Image Intensifier-Based Volume Tomographic Angiography Imaging System,”in Proc. SPIE Medical Imaging, vol.3032,pp. 612-619, 1997.
[11] J. Radon, "U¨ ber die Bestimmung von Funktionen durch ihre Integralwerte la¨ngs gewisser Manningfaltigkeiten," 1917, pp. 262–277.
[12] R. N. Bracewell, The Fourier Transform and Its Application, 2nd ed. : McGraw-Hill,Inc., 1986.
[13] Y. J. Roh, and H.S. Cho, “Implementation of Uniform and Simultaneous ART for 3D Reconstruction in An X-Ray Imaging system,” IEE Image Signal Process,Vol. 151, 2004.
[14] S. Kaczmarz, “Angenaherte Auflosung Von Systemen Linearer Gleichungen,” Bull. Acad. Pol. Sci. Lett. A, Vol. 6-8A, pp. 355-357, 1937.
[15] K. Tanabe, “Projection Method for Solving a Singular System,” Numer. Math., Vol. 17, pp. 203-214, 1971.
[16] A. H. Andersen, and A.C. Kak, “Simultaneous Algebraic Reconstruction Technique (SART): A Superior Implementation of The ART Algorithm,” Ultrasonic Imaging, Vol. 6, pp. 81-94, 1984.
[17] A. H. Andersen., “Algebraic reconstruction in CT from limited views”. IEEE Trans Med Imaging, Vol.8,pp. 50-55, 1989.
[18] M. Jiang, and G. Wang, “Convergence of the simultaneous algebraic reconstruction technique (SART),”IEEE Trans. Image Processing;,Vol.12,pp. 957-96, 2003
[19] NVIDIA. The CUDA homepage, http://www.nvidia.com/cuda
[20] A. Smith, “Full-field breast tomosynthesis,”Radiol Manage, Vol.27, pp. 25-31, 2005.
[21] K. Mueller, and R. Yagel, and J. J. Wheller, “Anti-Aliased Three-Dimensional Cone-Beam Reconstruction of Low-Contrast Objects with Algebraic Methods,” IEEE Transactions On Medical Imaging, Vol. 18, No. 6, June 1999.