Document block identification using a neural network

This paper describes a new method that clusters the content of a mixed type document in text or nontext areas. The proposed approach is based on a new set of textural features combined with a two stage neural network classifier. The neural network classifier consists of a principal components analyzer and a Kohonen self organized feature map. Document blocks are classified as text, graphics and halftones or to secondary subclasses corresponding to special cases of the primal classes. The proposed method can identify text regions included in graphics or even overlapped regions, that is, regions that cannot be separated with horizontal and vertical cuts. The performance of the method was extensively tested on a variety of documents with very promising results.