Classification of a Text and a Graphics Images with Asymmetrical Filters

Classification from text/graphics images is an important procedure for automatic digitization of documents. Based on the shapes in the basis image, the asymmetrical filters are developed in this paper. By filtering with the asymmetrical filters, the horizontal, vertical, and inclined edgespsila information are fetched from the input image. With applying a classified algorithm, the input image can be classified as a text image or a graphics image. Experiments are carried out for some samples to demonstrate the computational advantage of the proposed method.