Towards a unified framework for identity documents analysis and recognition

Identity documents recognition is far beyond classical optical character problems. Automated ID document systems are tasked not only with the extraction of editable and transferable data but performing identity validation preventing fraud, an increasingly high cost error. A significant amount research directed to creation analysis a specific focus for subset types, or particular mode image acquisition, however, one challenges modern world increasing demand from wide variety sources, such as scans, photos, video frames, well in virtually uncontrolled capturing conditions. In this paper, we describe scope context problem its challenges; analyze existing works on implementing systems; set task construct unified framework recognition, which would be applicable different types sources conditions, scalable enough support large number types. The aim presented serve basis developing new methods algorithms more heavy forensics, fully automated personal authentication fraud prevention.