Optimized robust multi-sensor scheme for simultaneous video and image iris recognition

Today, recognition of people by the iris is widely used when secure identification of a person is needed. Iris biometric identification systems should be able to work with heterogeneous iris images captured by different types of iris sensors. However, stable iris recognition systems that are effective for all types of iris cameras are not readily available. These systems should also be able to work simultaneously with images and video frames. In this work, we present an optimized robust multi-sensor scheme with a strategy that combines video frame quality evaluation with robust fusion methods at segmentation level for simultaneous video and image iris recognition. As part of the proposed scheme, we presented a Modified Laplacian Pyramid-based fusion method at segmentation stage. Experimental results on the Casia-V3-Interval, Casia-V4-Thousand, Ubiris-V1 and MBGC-V2 databases show that the optimized robust scheme increases recognition accuracy, and is robust to different types of iris sensors and able to simultaneously work with video and images.