車牌識別系統的設計與實現參考畢業論文-資料下載頁

　　

【正文】 數據是隨機取得的，幾乎是同樣數目的正例和反例被選入測試集。SVM 將會以懲罰參數 C=1 訓練各個不同子集的每個例子。使用不同內核和參數所獲得的結果如下所示。注意，為了實驗簡單，所有操作均在Matlab軟件上完成。 圖5 線性內核的ROC曲線 圖6 二次多項式內核的ROC曲線 圖7 當σ=1時，高斯RBF內核的ROC曲線 圖8 當σ=30時，高斯RBF內核的ROC曲線圖9 當σ=70時，高斯RBF內核的ROC曲線 圖10 當σ=3000時，高斯RBF內核的ROC曲線圖11 當a=1,b=1時，MLP內核的ROC曲線從圖511所示，我們可以發現當σ=70時，線性內核與高斯RBF內核一道可以取得幾乎完美的結果，而其他的方法卻都徹底的失敗，不是由于識別不出任何事物，就是存在大量分類錯誤，特別是假的正情況，但是仍然優于隨機分類方法。上述結果大多是從不具備足夠挑戰的數據中得到，MIT汽車數據庫內的圖像都非常類似，并且易于分類。在未來我們的工作里，為了在現實中能夠識別多種模型/品牌的汽車，我們將用更多區分度和準確性更高的特征來檢測更復雜的數據集。參考文獻[1] M. 1. Minsky and S. A Papert, Perceptrons. Cambridge, Ma: MIT Press, 1969, .[2] V. N. Vapnik, The Nature of Statistical Learning Theory, New York: SpringerVerlag, 1995, p 188.[3] V. N. Vapnik and A Ja. Chervonenkis, The necessary conditions for consistency of the method of empirical risk minimization. Yearbook of the Academy of Sciences of the USSR on Recognition, Classification, and Forecasting, vol. 2, pp. 217249, Nauka Moscow, 1989 (in Russian). English translation: Pattern Recogn. And Image Analysis, vol. I, no. 3, pp. 284305, 1991.[4] V. N. Vapnik and A. Ja. Chervonenkis, On the uniform convergence of relative frequencies of events to their probabilities. Rep. Academy Sci. USSR,pI81,n04,1968.[5] B. Boser, I. Guyon, and V. N. Vapnik, A training algorithm for optimal margin classifiers. in Proc. 539。h Annu. Wkshp. Comput. Learning Theory. Pittsburg, PA: ACM, 1992, pp. 144152[6] N. Dalal and B. Triggs, Histograms of oriented gradients for human detection. Computer Vision and pattern Recognition, 2005. IEEE Computer Society Conference on, vol. I, pp. 886893, June 2005.[7] V. Kecman, Support vector machines An introduction., StudFuzz 177, , 2005.[8] V. Kecman, Learning and Soft Computing, Support Vector Machines, Neural Networks and Fuzzy Logic Models, Cambridge, MA: MIT Press, 2001.[9] 1. C. Platt, Sequential minimal optimization: a fast algorithm for training support vector machines. , Technical Report MSRTR9814, April 1998.[10] 1. Mercer, Functions of positive and negative type and their connection with the theory of integral equation. , vol. A209, pp. 415446,1909.[11] T. Joachims, Making largescale SVM learning practicaL, in Advances in Kernel Methods Support Vector Learning, chapter II, Cambridge, Ma: MIT Press, 1999.[12] R. E. Fan, P. H. Chen, and C. J. Lin, Working set selection using second order information for training SVM, Journal of Machine Learning Research, vol. 6, , 2005.[13] , Distinctive image features from scaleinvariant keypoints. International Journal of Computer Vision, voL 60, no. 2, pp. 91110, 2004.[14] H. Bay, A. Ess, , and 1. Van Gool, SURF: Speeded up robust features. Computer Vision and Image Understanding voL 110, no. 3,pp. 346359,2008.[15] , J. Malik, and , Matching shapes. The 839。 ICCV, Vancouver, Canada, pp. 454461, 2001.[16] , , and D. Snow, Detecting pedestrians using patterns of motion and appearance. The 939。 ICCV,Nice, France, voLl, 741,2003.[17] C. Papageorgiou and T. Poggio. A trainable system for objet detection. International Journal of Computer Vision, , no. I, , 2000.附錄2 原文