基于Adaboost算法的虹膜合格状态检测

虹膜图像的采集是业界公认的难点，也是制约虹膜识别广泛应用的主要原因。怎样快速方便地采集到一幅清晰度足够并且有丰富纹理的高质量虹膜图像，对虹膜识别的速度和准确度都起着至关重要的作用。提出了一种基于Adaboost的虹膜图像合格状态检测和定位算法，能够快速有效地一次性检测出虹膜图像采集中的各种不合格图像，例如图像中没有完整眼睛、眼睛睁开程度不够、闭眼、斜视、运动模糊等。大量实验结果表明，该算法具有较好的检测准确率，对各种干扰情况有较强的鲁棒性，并且检测速度快，能够达到实时要求。对于检测合格的图像，还可以大致定位出虹膜在图像中的位置，为后续的虹膜定位节省时间。

Qualified status detection algorithm for iris image based on Adaboost

How to get qualified iris image is one of the difficulties in iris recognition. It also restricts the extensive application of iris recognition. Thus, how to quickly and easily collect an iris image with high resolution and rich texture plays a vital role for the speed and accuracy of an iris recognition algorithm. In this paper, a qualified status detection and locali-zation algorithm for iris image based on Adaboost is proposed, which can quickly and efficiently detect unqualified iris images, such as incomplete eyes, blinking eyes, closed eyes, squinting eyes, motion blur etc. A large number of experimental results show that the proposed algorithm has an accurate and fast performance. And it is also very robust for a variety of interference. In addition, for the qualified image, the proposed method can roughly locate the position of the iris, which increases the speed of the subsequent iris location.