Image recognition using a radial neighborhood method

Numeral character recognition is considered in this paper. Three recognition methods are proposed. The radial neighborhood method is basic, whereas the slice method and the method, based on a modified Hausdorff distance, are additional. A series of experiments is performed to compare these methods with the CL-approach and the correlation algorithm. Artificial and real noised images are used as the test samples. Resolution of these images is low. Experimental results reveal an effectiveness of the proposed methods, most notably the radial neighborhood method.