Morphological analysis and recognition of scanned raster nautical chart

In current practice in the development of hydrographic information systems, the automatic acquisition of spatial data has been considered to be very difficult. One of the key problems is recognition and correct separation of a feature from a complex graphic background. For these problems, this article presents a new concept of graph subsets to describe graph structures and composition rules of symbols in nautical charts. Based on this concept, morphological methods and operations in graph theory are unified into a coherent mathematical framework specially designed for analysis and recognition of binary chart images. In particular, an experimental system composed of feature extraction, symbol recognition, geometric reasoning, and intelligent control is developed for automatic digitization of nautical charts.