A Hierarchical Algorithm for Document-Images Fast Matching

As digital Libraries and document images largely use in the network, how to retrieve them quickly become one of the key issues. This paper presents a hierarchical matching algorithm to achieve fast retrieval of document images. Firstly, we can quickly find the possible matching location through approximate string matching algorithms, and then, use this location as a reference point in the target image, extract sub-block which has the same size as template image and compute the correlation coefficient between sub-block and template image. According to correlation coefficient values, we can accurately know whether the template image´s information exist in the target document, finally, experimental results demonstrate the feasibiLity of the algorithm.