Classification Of Table Form Documents Using High Order Correlation Method

Abstract

The recognition of table form documents is useful in office automation and file management. This paper presents a new approach for automatic document classification using high order correlation (HOC) method. HOC was originally used to recursively compute the cross-correlations between consecutive data in order to extract moving target tracks in three-dimensional (3-D) space. The most similar application in 2-D space is curve detection. A table form document consists of lines, characters and sometime graphs. It would be very convenient to use HOC to perform segmentation and classification on this type of images. The results contribute to many applications such as document identification and optical character recognition (OCR). It was shown that HOC could be implemented using a neural- network type of structure. This will greatly improve the efficiency of computation. The effectiveness of this approach will be demonstrated in the simulation results.