地面偵測跟空中偵測為現在主要的兩種偵測方法，地面偵測有環境限制、偵測範圍過小等缺點，所以地面偵測將會比空中偵測費時，空中偵測比較適合用在複雜的環境及大面積的監視。本文利用空中偵測來進行汽車物件的偵測。本系統是建立在機器學習的方法上，並分成訓練及測試兩部分。對於物件識別來說，樣本收集是最為重要的一部分，在此套系統中，主要是以汽車的擋風玻璃及車燈為主要特徵去進行識別並利用多個局部二值特徵去訓練樣本。透過影像處理去改善辨識率、誤判率及處理速度。從實驗結果顯示，此系統可以在高度變化不大的動態平台上運行整天。此系統可以架構在高度變化不大的UAV上，也可以發展成即時偵測系統。

The main surveillance methods are divided into two section, ground surveillance and aerial surveillance. Ground surveillance has disadvantages to subject to environmental constraint and smaller detection range to be more time consuming. Aerial surveillance is more suitable for complex environments and covers a much larger spatial area. This thesis develops an aerial surveillance on ground object detection using vehicle detection. The proposed system is constructed based on machine learning algorithm which involves training process and testing process. Sample collection is the most important part of object detection. Wind screen and lamps are the main characteristics chosen to recognize a vehicle in the proposed AdaBoost vehicle detection system. In this thesis, it trains samples uses multiple train cascaded LBP classifier. Through the image process, it can improve the detection rate, false alarm rate and process speed. The experiment results show the propose system can run all day on a dynamic platform with constant attitude to detect various vehicles. The proposed system can be applied on the UAV in constant altitude autopilot as well as any real-time surveillance applications.

[1] J. Y. Choi, Y. K. Yang, “Vehicle detection from aerial images using local shape information”, Advances in Image and Video Technology, vol. 5414, January 2009, pp. 227–236.
[2] S. Hinz, A. Baumgartner, “Vehicle detection in aerial images using generic features, grouping, and context”, Proceedings of 23rd DAGM Symposium Munich, vol. 2191, September 2001, pp. 45–52.
[3] X. Shi, H. Ling, E. Blasch, W. Hu, “Context-driven moving vehicle detection in wide area motion imagery”, 21st IEEE International Conference on Pattern Recognition (ICPR), November 2012, pp. 2512-2515.
[4] L.W Tsai, J.W Hsieh, K.C Fan, “Vehicle detection using normalized color and edge map”, IEEE International Conference on Image Processing, vol. 16, March 2007, pp. 850-864.
[5] X. Cao, C. Wu, P. Yan, X. Li, “Linear SVM classification using boosting HOG features for vehicle detection in low-altitude airborne videos”, 18th IEEE International Conference on Image Processing (ICIP), September 2011, pp. 2421-2424.
[6] S. Sahli, Y. Ouyang, D. A. Lavigne, “Robust vehicle detection in low-resolution aerial imagery”, International Society for Optics and Photonics, vol. 7668, April 2010, pp. 76680G-76680G.
[7] T. P. Breckon, S. E. Barnes, M. L. Eichner, K. Wahren, “Autonomous real-time vehicle detection from a medium-level UAV”, 24th International Conference on Unmanned Air Vehicle Systems, 2009, pp. 29.1-29.9
[8] C. John, “A computational approach to edge detection”, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PAMI-8, November 1986, pp. 679-698.
[9]P. Viola, M. J. Jones, “Robust real-time face detection”, International journal of computer vision, vol. 57, May 2004, pp. 137-154.
[10] R. Lienhart, J. Maydt, “An extended set of haar-like features for rapid object detection”, 2002 International Conference on Image Processing, Vol. 1, 2002, pp. I-900 - I-903.
[11] C. P. Papageorgiou, M. Oren, T. Poggio, “A general framework for object detection”, 1998 sixth international conference on Computer vision, January 1998, pp. 555 - 562.
[12] T. Ojala, M. Pietikäinen, T. Mäenpää, “Multiresolution gray-scale and rotation invariant texture classification with local binary patterns”, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 24, July 2002, pp. 971-987.
[13]Y. Freund, R. E. Schapire, “A decision-theoretic generalization of on-line learning and an application to boosting”, Journal of computer and system sciences, vol. 55, December 1997, pp. 119-139.