Thresholding and character recognition from a digital raster image

The basic aim of this paper is recognition of a character from a thresholded picture file, and converting it to text for further applications. The program first accepts a picture of raster type as input and removes the background noise from the picture. An iterative method of thresholding is presented which performs this operation. The picture is then passed through a number of pre-processing steps before recognition of characters. Character recognition using perceptual distance functions was first tried out. Because of the computational complexity involved in this method, a new method of character recognition using 2-dimensional discrete correlation is proposed in this paper. A comparison between the two methods is presented