查詢結果分析
來源資料
頁籤選單縮合
題名 | 線上手寫中文文字辨認系統的筆段抽取=Line Segment Extraction Methods in On-Line Chinese Character Recognition |
---|---|
作者 | 周國森; 張保忠; 徐克華; 陳玲芬; 許超智; 王亮盛; | 書刊名 | 電信研究 |
卷期 | 28:5 1998.10[民87.10] |
頁次 | 頁655-675 |
分類號 | 312.74 |
關鍵詞 | 線上手寫中文文字辨認系統; 筆段抽取; Stroke; Segment; OLCCR; On-line Chinese character recognition; DP; Dynamic programming; |
語文 | 中文(Chinese) |
中文摘要 | 特徵抽取問題一直是圖形與文字辨認中最重要的研究課題之一,對於中文文字的 字形結構而言,直線式的筆段是最基本的組成單元,也是最簡單而有效的特徵,因此以筆段 特徵為基礎的文字辨認系統可以自根本來解決文字辨認上所需面對的諸多問題,本文之內容 在於探討線上手寫中文文字辨認系統的筆段抽取方法,文中提出三種筆段抽取的方法:第一 種是遞迴式的筆段抽取法則,其將筆劃做首尾遞迴式搜尋可能的筆劃轉折點,並利用臨界角 及長短筆段判斷的判別方法以去除多餘而無意義的筆段,使中文文字的筆劃可以以最少的的 筆段予以表示。第二種方法是位置補償式的筆段抽取法,其對筆段的抽取採用角度方向做向 量式的相近角度吸收外,並根據中文文字曲線特性的筆段筆劃關係,對筆劃轉折點的位置做 補償式的對應,使中文文字筆劃轉換成簡單而具有規則筆段之組合。第三種方法則是以動態 規劃法來抽取輸入中文文字的筆段,其目地在於嘗試抽取式出最少的筆段數,以期使其與輸 入筆跡筆劃間的誤差達到最小,此一誤差量測可基於差值總合、最大差值與長度差值三種之 一。 |
英文摘要 | Finding an efficient feature extraction method is an important study in pattern/character recognition. Straight line segment is the simplest and the most efficient feature of Chinese characters. Straight line segment is the fundamental element of Chinese character, and the straight line segment feature satisfy the characteristics of Chinese character. In this paper, three segment extraction methods are proposed for On-Line Chinese Character Recognition (OLCCR). The first is the recursive segment extraction method, we find the possible cusps of a stroke by recursive procedure, and then remove the redundant cusps and short segments by way of absorbing critical angles and long-short decision rules. The second is the Positional Compensation segment extraction method which gets segments of a stroke by vectors and decide the cusps with position modulation. In the third extraction method, a dynamic programming algorithm which can extract the line segments from the loci of Chinese character in running hand or cursive writing is proposed. This algorithm tries to find the minimum line segments to fit the cursive stroke of character within the prescribed error bound. Three error measures are presented to construct a globally optimal approximation to the given stroke, including sun of deviation, peak deviation and length error. Extensive experiments are demonstrated to illustrate the performance of the proposed algorithms. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。