Text String Extraction in Delaunay Triangulation

This paper establishes a Delaunay triangulation model for detecting text strings in an image. In this method, text regions are represented based on Delaunay triangles that can be extracted easily. A searching mechanism based on the left and right arms of a discrete symmetric axis is developed to locate text in an image. The discrete symmetric axis consists of a group of off-curve-edge adjacent Delaunay triangles that are identifiable. Our experiment shows that the approach can detect multi-orientated texts and is efficient and effective. Key words: text extraction, connected component, Delaunay triangulation, and discrete symmetric axis

• Type: Conference
• Year: 2003
• Where: DICTA
• Authors: Huimin Grace Guo, Yi Xiao, Hong Yan
• Keywords: Delaunay Triangles, Delaunay Triangulation, DICTA 2003, DICTA 2007, Discrete Symmetric Axis