ABSTRACT
Currently using the iris texture region for biometric recognition makes the systems more accurate. The use of iris recognition systems in uncontrolled environments allows to have the real conditions of acquisition without the cooperation of the person; in this type of iris recognition systems the video is involved for the acquisition of the iris information, frames of the video sequence are acquired over long distances and the person moves at a normal pace of walking. However, the overall performance of the iris recognition system may be significantly reduced by operating with uncontrolled behavior of the users; biometric information suffers quality degradation and produce errors in different stages of the recognition process. An analysis of topics related to biometric iris recognition systems is carried out in this chapter. The focus here is on the development of mathematical methods to recognize the person through their iris using the images and video acquired in an uncontrolled environment. Some strategies to increase the performance of iris recognition system by including new stages for evaluation (i.e., image quality and segmentation failures). The problems, challenges and proposals for iris segmentation in poorly controlled environments are also discussed. Another issue addressed in the chapter is the problem of extracting iris features, including the new paradigm of deep neural networks. The characteristics of the main reference databases used to validate the developed iris recognition systems. Some of our main experimental results in the development of iris recognition systems in non-cooperative environments are analyzed. It is demonstrated that the development of iris recognition systems in uncontrolled conditions is a very topical issue that gives the wide spectrum of possibilities in real applications and can have lower cost than traditional systems, such as forensic identification, access controls to mobile devices (cell phones and tablets), criminal and police investigation, banking operations etc.
