Application of image recognition in GIS

In view of the existing issues on extraction of GIS vector data from maps, we propose an image recognition method based on wavelet transformation. Using the method, we first made the wavelet transformation of a color urban traffic map and took the value near zero points of the wavelet coefficient, and then, color model transformation was adopted to obtain a gray image. Afterwards, an appropriate threshold was chosen and the gray image was transformed into a bilevel image. In the end, the erosion algorithm in the mathematical morphology was used to thin the image in order to identify the road information. The experimental result showed that the method could be used to effectively identify the road information from an urban traffic map.