Intelligent interactive map recognition using neural networks

An intelligent interactive map recognition method using the case learning ability of neural networks as proposed. It effectively uses operator judgements about local segmentations of map figures as neural network learning data. With this method, the recognition accuracy can be improved and the number of interactive operations needed in the recognition process can be decreased through the operation processes. The method is implemented with a single recognition algorithm which has no dependence on the kind of map figure. Experiment results of applying this method to the recognition of contour lines, longitudinal/latitudinal lines and sidewalks in a mapdraw show improved accuracy and a decreased number of interactions.<
>