Allocation and clustering of text and graphic units on grayscale images

The method of analysis the documents presented in the form of images, containing as text (characters, digits) and graphics parts (pictures, photos) is considered. It is shown, that the tasks of allocation and clustering of the text and graphics information in similar documents can be solved, though available distinctions, with application of the same tools, including artificial neural networks (ANN). The analysis is considered as the first step of the common task of compound documents' clustering solution on the basic of ANN. The problem of documents processing technology presented in various formats is still urgent.