Map Background and Form Separation

Our goal is the extraction of semantic content from raster map images which may then be exploited in a variety of ways; e.g., the analysis of aerial imagery. The major map features that we want to extract include: roads and road intersections, texture, text, and miscellaneous symbols. Where appropriate these may be vectorized, and the image coordinates may be recorded as well. Meta-data may be provided including the type of map (e.g., USGS), any specific color encoding of features (e.g., USGS maps may use from 6 to 13 specific colors), geo-spatial information, and names of features may be known and sought in the image (e.g., rivers, lakes, roads, towns, etc.).