在交通基礎設施，如橋樑，隧道和高速公路的週期性結構健康檢查，是維護的根本且至關公共安全的任務。傳統上，藉由人的視覺檢查是最常使用的方法。它通常涉及到大量的人力資源，且可能由於錯誤、無知判斷使得較低效率和錯誤辨認的嚴重缺陷。受到谷歌街景服務的啟發，一個自動缺陷檢測系統，它集成了多個高畫質攝影鏡頭，GPS或任何位置識別機制，和圖形識別技術，依此可以實現和安裝在維護的車輛上，用來取得大型交通基礎設施的影像並且識別潛在的缺陷。該系統可以加強甚至取代了常規的視覺檢查。其中如何自動識別從圖像中的缺陷將是一個很大的挑戰。本研究論文提出了混凝土裂縫的檢測方法，它包括三個階段，（ㄧ）圖像前處理以及圖像補丁的特徵提取;（二）支持向量機分類器對25x25像素圖像補丁的裂縫檢測及（三）根據裂縫的嚴重程度給予照片排序。本方法透過接收者操作特徵（ROC）分析顯示排序結果的偵測率和假陽性率之間的權衡評估。其中ROC曲線下的面積（AUC）為性能的數值指標，且該面積始終是小於1。此提出的方法能夠在6秒鐘內處理60張792x633像素的照片且達到0.92的AUC值。結果顯示，該系統可以自動有效地檢測混凝土裂縫。

Periodically structural health inspection for the transportation infrastructure such as bridges, tunnels and freeways is a fundamental task for maintenance and crucial to the public safety. Traditionally, human visual inspection is the most utilized inspection method. It usually involves large amount of human resources, which can be inefficient and misidentifying important defects due to misjudgment and ignorance. Inspired by the Google Street View, an automatic defect detection system, which integrates multiple high quality cameras, a GPS or any location identification mechanism, and pattern recognition technologies, can be implemented and mounted on a maintenance vehicle to capture thorough images of large-scale transportation infrastructures and identify potential defects. The system can enhance or even replace the common visual inspection. How to automatically identify defects from the images will be a big challenge. This research proposed a concrete crack detection approach, which involves three stages, (i) image processing for feature extraction; (ii) SSVM classifier for small patch, 25x25 pixels, crack detection and (iii) ranking the photos based on the severity of crack. The proposed method evaluates the system by Receiver Operating Characteristic (ROC) analysis that shows the tradeoff between the detection rate and false positive rate of the ranking result. The area under the ROC curve (AUC) is a scale index for the performance that is always less than 1. This method is able to process sixty 792x633 pixels photos within 6 seconds and achieve a 0.92 AUC value. The result shows that the system can automatically detect concrete crack effectively and efficiently.

[1] Ikhlas Abdel-Qader, Osama Abudayyeh, and Michael E Kelly. Analysis of edgedetection
techniques for crack identification in bridges. Journal of Computing in
Civil Engineering, 17(4):255–263, 2003.
[2] Corinna Cortes and Vladimir Vapnik. Support-vector networks. Machine learning,
20(3):273–297, 1995.
[3] Luigi Di Stefano and Andrea Bulgarelli. A simple and efficient connected components
labeling algorithm. In Image Analysis and Processing, 1999. Proceedings. International
Conference on, pages 322–327. IEEE, 1999.
[4] Pedro Domingos. A few useful things to know about machine learning. Communications
of the ACM, 55(10):78–87, 2012.
[5] Yusuke Fujita, Yoshihiro Mitani, and Yoshihiko Hamamoto. A method for crack
detection on a concrete structure. In Pattern Recognition, 2006. ICPR 2006. 18th
International Conference on, volume 3, pages 901–904. IEEE, 2006.
[6] Atsushi Ito, Yoshimitsu Aoki, and Shuji Hashimoto. Accurate extraction and measurement
of fine cracks from concrete block surface image. In IECON 02 [Industrial
Electronics Society, IEEE 2002 28th Annual Conference of the], volume 3, pages
2202–2207. IEEE, 2002.
[7] Kwang In Kim, Keechul Jung, Se Hyun Park, and Hang Joon Kim. Support vector
machines for texture classification. Pattern Analysis and Machine Intelligence, IEEE
Transactions on, 24(11):1542–1550, 2002.
[8] Gary Klassen and Bill Swindall. Automated crack detection system implementation
in aran. In Digital Image Processing: Techniques and Applications in Civil Engineering,
1993.
[9] Yuh-Jye Lee and Olvi L Mangasarian. Ssvm: A smooth support vector machine for
classification. Computational optimization and Applications, 20(1):5–22, 2001.
[10] Long Liu and Guang Meng. Crack detection in supported beams based on neural
network and support vector machine. In Advances in Neural Networks–ISNN 2005,
pages 597–602. Springer, 2005.
45
[11] Zhiwei Liu, Shahrel A Suandi, Takeshi Ohashi, and Toshiaki Ejima. Tunnel crack
detection and classification system based on image processing. In Electronic Imaging
2002, pages 145–152. International Society for Optics and Photonics, 2002.
[12] Henrique Oliveira and Paulo Lobato Correia. Supervised crack detection and classification
in images of road pavement flexible surfaces. Recent Advances in Signal
Processing, pages 159–184, 2009.
[13] Nobuyuki Otsu. A threshold selection method from gray-level histograms. Automatica,
11(285-296):23–27, 1975.
[14] Peggy Subirats, Jean Dumoulin, Vincent Legeay, and Dominique Barba. Automation
of pavement surface crack detection using the continuous wavelet transform. In Image
Processing, 2006 IEEE International Conference on, pages 3037–3040. IEEE, 2006.
[15] Arian R Van Erkel and Peter M Th Pattynama. Receiver operating characteristic
(roc) analysis: basic principles and applications in radiology. European Journal of
radiology, 27(2):88–94, 1998.
[16] Vladimir Vapnik. The nature of statistical learning theory. springer, 2000.
[17] Pingrang Wang and Hongwei Huang. Comparison analysis on present image-based
crack detection methods in concrete structures. In Image and Signal Processing
(CISP), 2010 3rd International Congress on, volume 5, pages 2530–2533. IEEE,
2010.
[18] Tomoyuki Yamaguchi and Shuji Hashimoto. Fast crack detection method for largesize
concrete surface images using percolation-based image processing. Machine Vision
and Applications, 21(5):797–809, 2010.
[19] Tomoyuki Yamaguchi, Shingo Nakamura, and Shuji Hashimoto. An efficient crack
detection method using percolation-based image processing. In Industrial Electronics
and Applications, 2008. ICIEA 2008. 3rd IEEE Conference on, pages 1875–1880.
IEEE, 2008.
[20] Tomoyuki Yamaguchi, Shingo Nakamura, Ryo Saegusa, and Shuji Hashimoto. Imagebased
crack detection for real concrete surfaces. IEEJ Transactions on Electrical and
Electronic Engineering, 3(1):128–135, 2008.