由於現今人力的成本逐漸增加，自動化設備的需求也與日俱增，許多的企業廠商都希望以自動化設備取代傳統人力，達到增加產線效能並降低開發成本的目的。而在電路板檢測的領域，已有許多廠商推出多功能的檢測設備，皆採用工業等級的電腦為主要的運算系統，導致製造成本過高、設備體積龐大且維修不易的情況，但隨著數位訊號處理器(Digital Signal Processor, DSP)的發展，其具備運算效能高、體積小、成本低且消耗功率低等優點，適合用來取代工業電腦等高成本的運算系統。故本論文主要目的便是利用DSP做為自動光學檢測系統的運算核心，搭配GigE高解析攝影機來擷取影像資訊，透過DSP在複雜運算上的優異表現，準確的達到電路板的瑕疵檢測與判別。使用者可透過後方的人機介面進行系統的管控，除了可以介面上處理參數的設定，也可以觀看到檢測的結果。

Due to the gradual increase in the cost of workforce, demand for automatic equipment is also increasing. Many businesses and manufacturers are trying to develop automatic inspection equipment to replace inspecting by human. It could achieve the purpose of increasing efficiency of production and reduce the cost. With the development of Digital Signal Processor (DSP), DSP become faster, smaller, low-cost and low power. Replacing industrial computers and other high-cost operating system is suitable. Therefore, the purpose of this paper is to use DSP as main operating system of automatic optical inspection system. Combining with a GigE camera to capture the image and transfer image information to DSP by Ethernet. And because of DSP outstanding performance in complex operations, we can accurately inspect defects and quickly identify the type of defects. The results will be transferred to human-machine interface and users can get the information of PCB.

[1] W. Y. Wu, M. J. Wang and C. M. Liu, “Automated Inspection of Printed Circuit Boards Through Machine Vision,” Computers in Industry, Vol. 28, No. 2, pp. 103-111, 1996.
[2] Texas Instruments Inc., TMS320DM643x DMP Enhanced Direct Memory Access (EDMA3) Controller, March 2008.
[3] M. Moganti, E. Fikert, H. D. Cihan and T. Shou, “Automatic PCB Algorithms: A Survey,” Computer Vision and Image Understanding, Vol. 63, No. 2, pp. 287-313, 1996.
[4] W. A. Bentley, “The Inspection: An Automatic Optical Printed Circuit Board (PCB) Inspection,” Proceedings of the Society of Photo-Optical Instrumentation Engineers, Vol. 201, pp. 37-47, 1979.
[5] R. C. Gonzalez and R. E. Woods, Digital Image Processing, Prentice-Hall, New Jersey, pp. 116-123, 2002.
[6] A. Prati, I. Mikic, M. M. Trivrdi and R. Cucchiara, “Detecting Moving Shadows: Algorithms and Evaluation,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 25, No. 7, pp. 918-923, 2003
[7] Automated Imaging Association, GigE Vision 2.0 Specification, November 2011.
[8] 劉國偉, “基於GigE視覺之印刷電路板自動光學檢測系統,” 國立台灣科技大學電機工程系碩士論文, pp. 18-34, 2012
[9] Basler Vision Technologies, Installation and Setup Guide for Cameras Used with Basler’s Pylon API, February 2008.
[10] U. Lamping, R. Sharpe and E. Warnicke, Wireshark User's Guide : for Wireshark 1.7, 2011.
[11] Texas Instruments Inc., TMS320C6000 TCP/IP Network Developer’s Kit User’s Guide, June 2003.
[12] Basler Vision Technologies, The Elements of GigE Vision, May 2007.
[13] Basler Vision Technologies, Controlling Packet Transmission Timing with the Interpacket and Frame Transmission Delays on Basler GigE Vision Cameras Application Notes, July 2008.
[14] J. P. Serra, Image Analysis and Mathematical Morphology, Academic Press, pp. 115-130, 1982.
[15] N. Otsu, “A Threshold Selection Method from Gray-level Histograms,” IEEE Transactions on Systems, Man and Cybernetics, Vol. 9, No. 1, pp. 377-393, 1979.
[16] Texas Instruments Inc., TMS320DM643x DMP Inter-Integrated Circuit (I2C) Peripheral User’s Guide, March 2011.
[17] Spectrum Digital Inc., TMS320DM6437 Evaluation Module Technical Reference, December 2006.
[18] Catalyst Semiconductor Inc., CAT24C256 256-Kb I2C CMOS Serial EEPROM, February 2006.
[19] Spansion Advance Information, S29GLxxxN MirrorBitTM Flash Family, October 2003.