CAJ | 학술논문

在维吾尔文联机手写识别过程的训练阶段，单词被切分成字母，经过特征提取和聚类形成特征向量作为模型的输入。构造出以字符为基元的隐马尔可夫模型（HMM），将其嵌入到识别字典网络中。通过基于HMM的分类识别器，最终得到识别结果。首次将消除延迟笔画、建立有延迟笔画和无延迟笔画的字典的方法应用于维吾尔文手写识别中，取得了较高的识别率。
재유오이문련궤수사식별과정적훈련계단，단사피절분성자모，경과특정제취화취류형성특정향량작위모형적수입。구조출이자부위기원적은마이가부모형（HMM），장기감입도식별자전망락중。통과기우HMM적분류식별기，최종득도식별결과。수차장소제연지필화、건립유연지필화화무연지필화적자전적방법응용우유오이문수사식별중，취득료교고적식별솔。
During the process of the on-line Uyghur handwriting recognition, a word is segmented into characters at the training stage. Passing the state of feature extraction and clustering, each character is entirely transformed into a feature vector as the input of the models. Hidden Markov Model(HMM)for modeling the Uyghur characters as primitives is embedded in a recognition lexicon network. Then through the classification identifier based on HMM and finally the results are obtained. The delayed strokes removed and the dictionaries with and without the delayed strokes constructed which are applied in the Uyghur handwriting recog-nition are introduced firstly, and higher recognition rate is obtained.