Efficient layout analysis of ancient manuscripts using local features

Abstract

A binarization-free layout analysis method for ancient manuscripts is proposed, which identifies and localizes layout entities exploiting their structural similarities on the local level. Thus, the textual entities are disassembled into segments, and a part-based detection is done which employs local gradient features known from the field of object recognition, the Scale Invariant Feature Transform (SIFT), to describe these structures. As the whole entity cannot directly be inferred from the mere positions of the interest points, a localization algorithm is needed that expands the interest points according to their scales and the classification score to regions that encapsulate the whole entity. Hence, a cascading algorithm is proposed that successively rejects weak candidates applying voting schemes.<br />Layout analysis is the first step in the process of document understanding; it identifies regions of interest and hence, serves as input for other algorithms such as Optical Character Recognition.<br />Moreover, the document layout allows scholars to establish the spatio-temporal origin, authenticate, or index a document. The evaluation shows that the method is able to locate main body text in ancient manuscripts. The detection rate of decorative entities is not as high as for main body text but already yields to promising results.