Abstract | ||
---|---|---|
In this paper, we present a new approach of printed text characterization with a statistical analysis of texture. This method is based on a visibility and complexity criterion of type font families. The rent is analyzed through its typographic form. So, we propose to label different kinds of text according to their visual aspect and their textural contents (especially their size, the line and letter spacing but also their complexity and their density). The texture is characterized with a statistical analysis based on samples randomly taken. This analysis is at the basis of the text labeling. It is dedicated to the classification of texts according to their eye-catching properties The global aim of this work is to recover the logical structure of document according to the relative importance of font-types. This discrimination has been realized by the use of a scale of legibility, complexity, and of structural relief of forms. |
Year | DOI | Venue |
---|---|---|
1998 | 10.1109/ICPR.1998.711390 | ICPR |
Keywords | Field | DocType |
visual criteria,printed test featuring,labeling,surface texture,image classification,statistical analysis,graphics,image texture,complexity,visual system,entropy,text analysis | Graphics,Legibility,Computer vision,Visibility,Text mining,Pattern recognition,Computer science,Image texture,Font,Artificial intelligence,Structure (mathematical logic),Contextual image classification | Conference |
ISSN | ISBN | Citations |
1051-4651 | 0-8186-8512-3 | 1 |
PageRank | References | Authors |
0.36 | 0 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Veronique Eglin | 1 | 12 | 2.85 |
Stephane Bres | 2 | 6 | 2.18 |
Hubert Emptoz | 3 | 383 | 38.09 |