應用電腦視覺技術於注音符號辨識

Computer Vision Techniques for Hidden Conditional Random Field-based Mandarin Phonetic Symbols I Recognition

酈頤芳(Yi-Fang Li)

指導教授 ： 李建誠

元智大學/電機通訊學院/通訊工程學系/碩士(2011年)

https://doi.org/10.6838/YZU.2011.00307

摘要

本論文提出一種應用電腦視覺技術於手寫國語注音符號第一式(Mandarin Phonetic Symbols I, MPS1)辨識的方法，此方法是基於隱藏式條件隨機域(hidden conditional random field, HCRF)來建立模型。HCRF是由條件隨機域(conditional random field, CRF)的結構加入隱藏變數而形成。此方法的優勢主要在於避免傳統上使用隱藏式馬可夫模型(hidden Markov model, HMM)而造成限制。本研究利用14維的特徵值，對每一個MPS1符號建立HCRF模型。本實驗的辨識率達 98.3784%。

關鍵字

手寫辨識 ； 注音辨識 ； 電腦視覺

並列摘要

This paper presents a handwritten recognition method for Mandarin Phonetic Symbols I (MPS1). The method is based on a hidden conditional random field (HCRF) model, which is an extension of a conditional random field (CRF) framework that incorporates hidden variables. The main advantage of the proposed method is that it avoids limitations of the traditional hidden Markov model (HMM)-based methods. This research built an HCRF model for each symbol in MPS1 and used fourteen-dimensional features. The recognition rate achieved 94.05%.

並列關鍵字

handwritten recognition ； handwritten

參考文獻

[1] J. Lafferty, et al., "Conditional Random Fields: Probabilistic Models for Segmenting and Labeling Sequence Data," in Proc. International Conference on Machine Learning 2001.

[5] A. Gunawardana, et al., "Hidden Conditional Random Fields for Phone Classification," in European Conference on Speech Communication and Technology, 2005.

[6] S. B. Wang, et al., "Hidden Conditional Random Fields for Gesture Recognition," in Proc. IEEE Computer Society Conference on Computer Vision and Pattern Recognition, New York, 2006, pp. 1521-1527.

[7] A. Quattoni, et al., "Conditional Random Fields for Object Recognition," in Neural Information Processing Systems Conference, Vancouver, 2004.

[9] Y. Liu, et al., "Hand-Gesture Based Text Input for Wearable Computers," in Proc. IEEE Computer Vision Systems, New York, 2006, pp. 8-13.