查詢結果分析
來源資料
頁籤選單縮合
題名 | Handwritten Character Recognition Using Nonlinear Normalization Algorithm and Fuzzy Nearest Prototype Classifier=利用非線性正規化與模糊最近原型分類之手寫字元辨識 |
---|---|
作者 | 陳兩嘉; 沈鴻光; | 書刊名 | 中原學報 |
卷期 | 23:1 1995.01[民84.01] |
頁次 | 頁113-121 |
分類號 | 312.84 |
關鍵詞 | 手寫字元辨識; 非線性正規化; 模糊最近原型演算法; Handwritten character recognition; Nonlinear normalization; Fuzzy nearest prototype algorithm; |
語文 | 英文(English) |
中文摘要 | 在本篇論文中,我們提出一個做手寫字元辨識的方法。使用吾人所提案之非線性正規法將每一細線化後之字元圖像模式化,成此模式為基礎,將得到四個方向的非線性正規化後之子圖像,而該四個子圖像的統計值即為此字元圖像的特徵。用此方法可提升對宇元因位移與大小變化之辨識能力。在每一手寫字元的特徵抽取後,吾人以模糊最近原型演算法去做分類工作。原型可更新,更新後的原型將用於分類器,而比自適性分類架構將進一步提升1.3%辨識率,總平均辨識率為98.5%。吾人提出之方法已由所作的實驗驗證。 |
英文摘要 | In this paper, we propose an approach to perform handwritten characters recognition. In our metohd, the pattern of each character after thinning is modeled using our proposed nonlinear normalization algorthm. Based on this model, we ues four directiona skeleton detection to get four subpatterns. The statistic values of the subpatterns are used as the feature sof the pattern. This method can improe the recogition rate wigh location shift and arbitrary size of an input character/ Afer features extraction of each handwritten character. we then use Fuzzy Nearest Prototype Algorithm for classificatron task. The prototype is updated. The updated prototype is used as classifier. This adaptive scheme of classifier increases recognition rate by 1.3%. The total average recogintion rate is 98.5%. Experiments have been conducted to verify the proposed method. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。