Automatic Line Network Extraction from Aerial Imagery of Urban Areas through Knowledge Based Image Analysis
Final technical rept. Nov 1986-Aug 1989
FORSCHUNGSINSTITUT FUER INFORMATIONS-VERARBEITUNG UND MUSTERERKENNUNG ETTLINGEN (GERMANY F R)
Pagination or Media Count:
Different methods for automatic detection of line objects applied to aerial images to extract streets from urban scenes are investigated. First, test results achieved from two existing methods of low level iconic image processing by stream following line tracking and structured parallel operations image filtering, feature extraction are given. Second, a medium level iconic image processing method developed for edge and area segmentation is described and results from image segmentation are presented symbolically. Then two preliminary approaches of high level symbolic processing by knowledge based blackboard oriented structure analysis are tested. One is originating with preprocessing by low level edge filtering, the other by medium level area segmentation. First results from the image understanding method for street network extraction are presented. Keywords Geodesy, Images, Optics, Artificial intelligence, Automation, Line network extraction, Aerial imagery of urban areas, Mapping, Extraction of line objects, Phoenix, Bietigheim.