文字集団の印字品質の数量表現

概要

論文の詳細を見る
This paper is concerned with print quality evalution of a large number of data. A computer simulation of evaluation for actual data shows that distributions of evaluation values do not obey the Gaussian distributions in the majority of cases. "Representative" which expresses a statistical value for a set of printed images, is proposed. The representative R is dfiened as: R=M/N, where M and N denote the 2nd moment about the origin and a total number of data, respective. Foer kinds of values of respectative(darkness, strokewidth, noise factor, and centroid deviation) are strongly correlated each other. The quality for a set of priented characters can be expressed by any one kind of representative.
一般社団法人情報処理学会の論文
1977-03-15