頁籤選單縮合
題名 | 中文點陣字形要素對字形閱讀視認性的影響研究=A Comparative Study on the Descriptors of Dot-Matrix Image of Chinese Characters |
---|---|
作者 | 游萬來; 蔡登傳; 陳文誌; | 書刊名 | 國立雲林技術學院學報 |
卷期 | 6:1 1997.01[民86.01] |
頁次 | 頁17-26 |
分類號 | 312.74 |
關鍵詞 | 中文點陣字形; 字形視認性; 字形描述法; 投影值譜幅; Dot-matrix images; Discriminability of Chinese characters; Descriptors; Projection profile; |
語文 | 中文(Chinese) |
中文摘要 | 可以描述中文點陣字形的方法甚多。為了求得一種可以預測評估中文視認性的有 效描述法,本研究首先探討各種可量化描述字形的方法,分別為筆畫的密度、筆畫數、對稱 度、邊界量及投影值譜幅( projection profile )。而投影值譜幅是屬於次數分布圖的性 質,可進一步計算求出譜幅的峰數、長度、對稱度、平均數、峰度及偏態等。本研究接著選 取最常用的198個中文字,就其較常見的明体24×24點陣字形分別以上述各種字形描 述法計算其描述值,進而以120名受試者量測這些中文字在電腦螢幕上可視認的最小視角 。實驗結果顯示:字形視認視角的大小與本研究所用各種中文字形描述法中的邊界量最為相 關(γ=0.84);換言之,邊界量最具有中文字形視認性的預測能力。另外視認視角的 大小和其筆畫的邊界量、筆畫的密度、筆畫數、投影值譜幅的峰數、長度等描述值呈正相關 ,而和其投影值譜幅的偏態、峰度及點陣對稱度等要素的描述值則呈負相關。此外,本研究 也求出了與視認視角相關性較高的各項描述值的迴歸曲線圖。本研究的結果在理論上可以用 來預測推估中文的閱讀視認性,在實用上則可用來評估字形設計的優劣。 |
英文摘要 | This study aims to gain understanding of the dot-matrix image factors which affect the discriminability of Chinese characters. Several quantifiable descriptors have been proposed for describing the dot-matrix image of Chinese characters. Among them are: image density, number of strokes, image symmetry, edge amounts, and projection profile. The projection profile, being a distribution of frequencies, can be further calculated for its number of peaks, length, symmetry, average, kurtosis, and skewness. The quantitative values of the above mentioned descriptors were then calculated for the most frequently used 198 Chinese characters. The smallest discriminable visual angles of the same set of characters were also tested on CRT with 120 subjects. The main discoveries of the results are as follows. The smallest discriminable visual angle of a Chinese character is positively related to the values of edge amount, image density, number of strokes, number of peaks in the projection profile, and length of the projection profile; however, it is negatively related to the values of skewness and kurtosis of the projection profile and image symmetry. Among the proposed descriptors, the edge amount has the highest value of coefficient of correlation(r=0.84)with the smallest discriminable visual angle. In other words, edge amount is the best descriptor to predict the discriminability of a Chinese character. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。