軌道運輸系統為主要的交通運輸工具，而固定鋼軌的彈性扣件常常因為連續鬆脫斷裂，造成列車出軌意外，因此對行車安全是相當的重要。本文利用電腦視覺(Computer Vision)技術開發一套數位影像辨識系統，應用於檢測捷運系統的軌道彈性扣件，可以取代目前以人力步行方式。此系統包括前端處理、鋼軌定位、扣件區域搜尋、扣件擷取、扣件辨識、螺栓定位、標線擷取及螺栓辨識八部份。前端處理是把彩色影像轉成灰階影像和消除雜訊，鋼軌定位使用灰階的變化特性進行鋼軌位置確認，扣件區域搜尋使用小波轉換(Wavelet Transform,WT)的理論進行搜尋，扣件擷取包括尋找適合的門檻值、形態學運算及物件搜尋的影像處理技術，扣件辨識經由扣件本身的結構特徵進行處理，螺栓定位和標線擷取是找出螺栓區域並將螺栓上的標線影像擷取出，螺栓辨識是利用標線的特徵進行處理。最後實際測試可以成功的辨識出正常或斷裂的扣件影像及辨識出螺栓是否鬆脫的影像，驗證本論文開發的系統具可行性。

Railway transportation system is an important transportation tool. Therefore, driving safety is of concern. However, accidents occur when a train goes off track due to broken elastic rail clips in the fixed rail. This research presents the development of a computer visual recognition system to detect the state of the elastic rail clips . This visual recognition system can be used in mass rapid transit systems to reduce substantial need of manpower for checking elastic rail clips at present. The visual recognition system under current development includes eight parts: preprocessing, orientation of the rail, search of elastic rail clip regions, selection of elastic rail clips, recognition of broken elastic rail clips, orientation of bolts, selection of reticles and recognition of loosen bolts. The preprocess transforms the colored images into gray-level images and eliminates noises. The orientation of the elastic rail clip uses characteristics of the gray-level variation to confirm the rail position. The search module uses wavelet transformation to search for elastic rail clip regions. Selection of a broken elastic rail clip depends on a suitable threshold, obtained by using techniques from morphological processing, object search, and image processing. The recognition of elastic rail clips is based on characteristics and structures of elastic rail clips. The orientation of bolts is used to look for bolt regions. Selection of reticles is based on reticle images on bolts. The recognition of a bolt is based on characteristics of the reticle. Experimental testing demonstrates the ability of this system to differentiate normal elastic rail clip images or broken elastic rail clip images and whether a bolt is loosened or broken. This result confirms the feasibility to develop such a visual recognition system.

[1]鄭國雄、張思，軌道工程，大中國圖書公司，民國八十八年三月。
[2]鐘國亮，影像處理與電腦視覺，東華書局，民國九十一年。
[3]A. K. Jain, Fundamentals of Digital Image Processing, Prentice-Hall, New York, 1989.
[4]M. Sonka, V. Hlavac, and R. Boyle, Image Processing, Analysis, and Machine Vision, 2nd Edition, PWS, New York, 1999.
[5]I. H. Witten, A. Moffat, and T. C. Bell, Managing Gigabytes: Compressing and Indexing Documents and Image, Van Nostrand Reinhold, 1994.
[6]李嘉雯，「車牌辨識系統」，國立臺灣科技大學碩士論文，民國八十九年。
[7]S. G. Mallat, “A Theory for Multiresolution Signal Decomposition: The Wavelet Representation,” IEEE Trans. on Pattern Analysis and Machine Intelligence, Vol.11, No.7, pp. 674-693, July 1989.
[8]C. S. Burrus, R. A. Gopinath, and H. Guo, Introduction to Wavelets and Wavelet Transforms: A Primer, Prentice-Hall, New Jersey, 1998.
[9]N. Otsu, “A Threshold Selection Method from Gray-Level Histogram,” IEEE Trans. Syst., Man, and Cybernetics, Vol.SMC-9, No.1, pp.62-66, 1979.
[10]R. C. Gonzalez, and R. E. Wood, Digital Image Processing, Addison-Wesley, New York, 1992.
[11]吳成柯、戴善榮、程湘君、雲立實，數位影像處理，儒林圖書，民國九十年。
[12]林家瑋，「眼位追蹤系統」，國立中山大學碩士論文，民國九十二年。
[13]R. Parisi, E. D. Di Claudio, G. Lucarelli, and G. Orlandi, “Car Plate Recognition by Neural Networks and Image Processing,” Proceedings of the IEEE International Symposium on Circuits and Systems, Vol.3, pp.195-198, 1998.
[14]T. C. Chen, and K. L. Chung, “An Efficient Randomized Algorithm for Detecting Circles,” Computer Vision and Image Understanding, Vol.83, pp.172-191, August 2001.