摘要： 在计算机辅助虹膜诊断中，虹膜卷缩轮的定位是其他虹膜异常特征检测的基础。针对该特征，提出一种基于图像梯度极值的虹膜卷缩轮检测方法。根据虹膜图像中卷缩轮及其附近区域在竖直方向上的灰度差异性，在部分归一化图像中，利用改进的垂直梯度算子对图像进行处理，提取竖直方向上的梯度极值点得到卷缩轮轮廓，并将轮廓点还原显示在原始灰度图像中。实验结果表明，该方法的定位准确率可以达到92.5%，具有较好的检测效果。
关键词:
图像处理,
灰度图像,
虹膜诊断,
虹膜卷缩轮提取,
灰度,
梯度,
图像归一化
Abstract: In the computer aided iris diagnosis, iris frill positioning is the foundation of the other iris anomaly feature detection. According this feature, this paper puts forward a detection method based on image gradient extreme. According to the vertical gray difference between the iris frill and its surrounding areas, it uses the improved gradient detection operator to get the image gradient value on the normalized image, extracts the maximum gradient value and gets the frill outline. Contour points are reduced and displayed in the original gray image. Experimental results show that the location accuracy of this method can reach 92.5%, and it has a good effect in the iris frill positioning problem.
Key words:
image processing,
gray image,
iris diagnosis,
iris frill extraction,
gray scale,
gradient,
image normalization
中图分类号: