Segmentation of text from color map images

Traditional maps contain information about land usage, transportation networks, political boundaries, topographic and hydrographic features, and annotations. The rapid growth in the volume and the complexity of maps has led to the development of computerized geographical and land information systems (GIS and LIS) to facilitate information management. Computerized representation of data contained in maps facilitates efficient searching and updating of information in such systems. Such a representation is also useful for creation of specialized maps to serve a particular need. However the conversion of existing paper-based maps into computer-readable databases is a time consuming and tedious task. This paper addresses one particular aspect of this task, viz., the segmentation of text from other graphical objects in digitized images of color maps