Multimedia Documents Segmentation and Classification

Abstract

Document page segmentation and classification are important parts of the document analysis process. Page segmentation is that the page is decomposed into blocks. It is also called “page decomposition” or “zoning”. The goal of page classification is to label these blocks according to their contents. In this paper, multimedia documents segmentation and document’s features classification is presented. Multimedia documents usually consist of a mixture of text, graphic, and image. One-scan run-length smearing algorithm with block merging is emphasized for document segmentation task. This smearing algorithm is a document page segmentation algorithm using a bottom-up approach. Document classification task is performed based on features of text, graphic and image. Separation and classification of text, graphic, and image are advantageous in reproducing, transmitting storing the multimedia document and extraction different parts of document.