Processing of Words Labels in Scanned Map Based on Singularity Detection

The words labels in the scanned map should be removed in some application. In this paper singularity detection based on wavelet is adopted to separate the words labels from the scanned map. One row is extracted from the scanned map, and then is convoluted with wavelet in different scales. Compared to the end of words labels, the words labels themselves have different Lipschitz exponents. In this paper the local minimum of Lipschitz exponents is proposed to separate the words labels from the background.