基於支持向量機之手寫辨識系統

A Support Vector Machine-Based Handwriting Recognition System

武越雄(Vu VietHung)

指導教授 ： 徐元寶

國立虎尾科技大學/電機資訊學院/資訊工程系碩士班/碩士(2016年)

摘要

本文的目的是實作一款以支持向量機(Support Vector Machine)為基礎之越南字母手寫辨識系統。用戶在開發的系統上寫入一個符號,系統會識別它，並給出建議字母的可能名單。我們使用方向梯度（Histogram of Gradient）進行特徵提取的直方圖，以及以徑向基函數（Radial Basis function）解決內核非線性之SVM分類。所得系統能達到90％的辨識率.

關鍵字

支持向量機 ； 方向梯度 ； 徑向基函數

並列摘要

The aim of this thesis is to implement a Support Vector Machine (SVM) on an Android cell phone to recognize handwritten Vietnamese alphabets. We use the Histogram of Oriented Gradients (HOGs) for feature extraction, and solve the non-linear SVM classifiers with Radial Basis Function (RBF) kernels. The user writes a character on the developed system, triggering the system to recognize it to output a possible list of suggestion alphabets. The resultant system recognizes the handwritten characters with an overall accuracy of 90%.

並列關鍵字

Handwritten recognition ； RBF ； SVM

參考文獻

[1] Bandera, A., Urdiales, C., Arrebola, F., and Sandoval, F., “2d object recognition based on curvature functions obtained from local histograms of the contour chain code,” Pattern Recognition Letters, Malaga, Spain, vol. 20, pp. 49 – 55, 1999.

[2] Belongie, S., Malik, J., and Puzicha, J., “Shape matching and object recognition using shape contexts, ”IEEE Transactions on Pattern Analysis and Machine Intelligence, Washington DC, USA, vol. 24,pp. 509–522, 2002.

[3] Bensefia, A., Nosary, A., Paquet, T., and Heutte, L., “Writer identification by writer's invariants,” Proceedings of the Eighth International Workshop on Frontiers in Handwriting Recognition, IEEE Computer Society, Washington DC, USA, vol. 15,pp. 274–279, 2002.

[5] Arazi, B., “Automatic handwriting identification based on the external properties of the samples,” Man and Cybernetics, IEEE Transactions of Systems, Baton Rouge, LA, USA, vol. 13, pp. 635–642, 1983.

[6] Vapnik, V., The Nature of Statistical Learning Theory, Springer, New York, 1995.