Intelligent Recognition And Research On Geographic Elements In The Scanned Map

The extraction and recognition of the geography information is the basis of geographic information system (GIS) and global position system (GPS). Automated vectorization has transcendent advantages and has become the main method to capture the digital information.In this paper, a method to recognize figure objects in the scanned map is put forward which is called an algorithm based on the sparse pixel without direction, after deeply studying and analyzing the representative vectorization methods. Using this method an automated vectorization system - Auto Vector is also established on the Visual C++6.0 environment.The steps of the new algorithm are to find a steady point, judge tracing direction, pick up the key points, get medial axis line, erasure the original raster line information, enter the lines web processing, at last reconstruct the figures. This method has the superiority over other methods, because it has no time-consuming processing such as thinning, contour tracing, wire's coding. Rather than visiting all the points through the wire's black area, the method visits less points with jumping sampling. It reduces most of datum to storage compared with the other methods based on the every pixel. The anti-jamming capability is improved obviously. A erasing process which can keep the intersection information avoids re-recognition and reduce the complexity. In order to avoid advancing tracing end in the intersection, an intersection disposal processing is added to the algorithm. It provides more facility for the whole map topology structure.In order to recognize and capture figures, minimal-angle principle and inertia principle are applied in this paper to realize polygons and fold lines' picking up. Otherwise, the author applies step arc vectorization algorithm, direction statistic distance morphology ,noise-removing algorithm and some basic operation towards binary images. The system built can reestablish the figures' original shape faster and better, and get an excellent result after vectorization.At present the research about vectorization is mainly among the engineering image. The method designed in this paper realizes the interface with Maplnfo which is a GIS software and satisfies the demand of GIS. It has theory meaning and application value in GIS.