A Method for Color Raster Map Annotation Recognition

Yilang SHEN,Tinghua AI,Rong ZHAO
DOI: https://doi.org/10.13203/j.whugis20150381
2018-01-01
Abstract:The annotations in raster maps contain important geographical location information,but the traditional recognition methods of raster map annotation separate annotations from geographic elements,making it difficult to deal with curved and sloping annotations.This paper proposes a method for color raster map annotation recognition based on geometric features.The method considers the orientation,size,density,and neighborhood of the connected components to extract annotation pixels,then corrects the direction by extracting the centerline of geographical elements,clustering,and rearranging annotations before finishing the recognition.We use GoogleMap in experiments,the proposed method was effective.
What problem does this paper attempt to address?