Cartographic Feature Extraction on ETL's (Engineer Topographic Laboratories') DIAL (Digital Image Analysis Laboratory) System

摘要

A number of feature extraction techniques have been developed and tested on the Digital Image Analysis Laboratory (DIAL) at ETL. These techniques include stereo matching, edge detection, texture extraction, and statistical pattern recognition on digitized aerial imagery, as well as interactive binary image cleansing (raster processing) on the derived results. Experiments have shown that each of these techniques has some limited success at performing its task, but as isolated entities, they fail to perform at a level needed in a cartographic feature extraction system. This paper discusses some of the capabilities of DIAL and concludes that a method of coordinating simple feature extraction techniques under the management of a heuristic rule-based system should be investigated.