Finding text Stroke width variety in city maps

Abstract

The Stroke width is an important and stable feature to describe the texts in the document images. In this paper, we propose a method for finding stroke width variety in city map images. Since in city maps the graphics lines and text labels are usually overlap with each other, it is difficult to find the stroke width in such images. On the other hand, texts are printed in a variety of widths. Knowing the major text stroke width is a prior knowledge before map processing like text extraction from graphics lines. In the proposed method, we find the candidate connected components that have significant stroke-width information. Then we locally assign a minimum stroke width to each pixel. For each candidate component, stroke width is determined. By clustering stroke width of components, we find major stroke widths. The experimental results on several varieties of city maps are reported and shown to be promising.