Page segmentation and classification based on pattern-list analysis

In this paper, a new algorithm based on pattern-list analysis is proposed for page segmentation and classification. There are three steps in the algorithm: the bounding rectangle location, the pattern formation and the pattern classification, after which the patterns that may be wrongly classified are further classified by their contextual information. Experimental results show the accuracy of the algorithm in segmenting text and non-text regions, especially for the case of document images with irregular-shaped halftone regions. The algorithm is valid only for binary document images.