基于双目视觉和旋转云台的远距离虹膜识别系统

 虹膜识别技术是一种新兴的生物特征识别技术，相对于指纹识别和人脸识别具有更高的准确率和稳定性，影响虹膜识别的瓶颈问题在于虹膜图像获取不太方便。当前该领域的热点就是研发一种可以在远距离、全自动地拍摄虹膜图像的设备，以解决虹膜识别的易用性问题。设计了一种基于双目视觉和旋转云台的远距离虹膜识别系统，该系统首先对摄像机进行标定，对双目摄像机中的人脸图像进行配准，从而估计人脸在空间中的大致位置；然后，系统利用旋转云台将虹膜摄像机自动对准人的眼睛并进行自动对焦，在捕获到清晰地虹膜图像后进行虹膜识别。实验证明，本文的立体视觉方法可以比较准确地定位物体空间位置，系统自动获取的虹膜图像质量好，识别率高。这种系统有望能够在未来的虹膜识别领域得以大规模应用。

Iris Recognition System at a Distance Based on Stereo Vision and Pan-tilt-zoom Camera

Iris recognition is an emerging biometrics technology. It is more accurate and robust than fingerprint recognition and face recognition, but with a bottleneck of the inconvenience in iris image acquisition. Many recent studies focus on the automatic iris recognition at a distance to make the operation more convenient. In this study, an iris recognition system is designed at a distance based on a stereo vision and a pan-tilt unit. The stereo vision system is pre-calibrated. After the human face detection, the system makes a face image registration between two stereo-vision cameras and uses the corresponding relationship to estimate the accurate spatial position of human eyes. The pan-tilt unit moves an iris camera towards the eye position, and with the auto-focusing function, the iris camera acquires high-quality iris images for the iris recognition. Experiments show that the stereo vision is accurate enough to estimate the spatial position and the acquired iris images are clear enough for a high performance of the iris recognition. The iris recognition system has a potential large-scale application in the future.