On the Segmentation of Color Cartographic Images
One main problem in image analysis is the segmentation of a cartographic image into its different layers. The text layer is one of the most important and richest ones. It comprises the names of cities, towns, rivers, monuments, streets, and so on. Dozens of segmentation methods have been developed to segment images. Most of them are useful in the binary and the gray level cases. Not to many efforts have been however done for the color case. In this paper we describe a novel segmentation technique specially applicable to raster-scanned color cartographic color images. It has been tested with several dozen of images showing very promising results.