基于Camshift算法的虹膜实时跟踪

针对人眼虹膜跟踪中存在的眨眼和眼睑遮挡问题,提出了基于Camshift算法的虹膜跟踪方法。首先使用Adaboost学习算法进行人眼初步定位,然后加入三庭五眼的比例模型精确定位人眼;在人眼定位的基础上使用一个圆形滑动窗遍历人眼灰度图像,其中平均灰度最小的圆形窗可初步定为虹膜区域;将以上步骤检测出的虹膜作为Camshift算法初始模板,建立虹膜的颜色概率图,利用虹膜的颜色特征完成跟踪。实验证明本算法的虹膜跟踪准确率达到84%;并能解决眨眼和眼睑遮挡问题,保证了跟踪过程的鲁棒性。

Iris Tracking Algorithm Based on Camshift

Aiming at the problem of the blink of an eye and eyelid occlusion for iris tracking systems, this paper presents a Camshift algorithm for iris tracking. Adaboost method is used to initially detect human eye. Then model of facial scale is used for precise positioning of the human eye. On the basis of eye location, a circular sliding window traverses the gray image of the eye. The area which is the lowest average gray intensity of sliding window is considered as the iris candidate area. Finally, area of iris is used as the initial Camshift template for tracking. Color information of the template is used for the next image sequences. The experiment results show that the proposed approach achieved the accuracy of 84%. The results demonstrate the algorithm can track the iris when the eyelid is occluded or eye is blinking.