Using Intersection Graphs for Smartphone-Based Document Localization

Abstract

This article is devoted to analyzing document localization in images and evaluation of the performance of mobile applications. The analysis is used to propose a new algorithm of document-image capture. The algorithm consists in determining segments of document boundaries and building an intersection graph that complies with a projective rectangle model. According to the evaluation of the performance of the algorithm, its document-localization efficiency is as high as 95% and it outperforms all the reviewed algorithms used in mobile applications.