提出了一种新的内侧指横纹识别方法.首先,对图像采集设备进行改进,在采集过程中固定了手指方向,使获得的指横纹感兴趣区域(ROI)之间只存在微小的平移变换,有利于提高图像匹配的精度.另外,在预处理阶段,利用Gabor滤波法检测手指线特征以分割手指,并从中提取出ROI.在特征匹配阶段,提出了一种利用投影比较进行定位的图像匹配方法,对ROI特征图像进行水平和垂直方向投影,通过比较所得的一维向量实现感兴趣区域的精准定位.评估系统建立在包含来自于77个人的820幅图像的数据库上,等错误率仅为0.61%,单次匹配时间为3.1ms,证明该算法可快速实现指横纹特征识别,准确率较高.
A novel personal recognition method based on the inner-knuckle-print is proposed in this paper.A data acquisition device is developed to capture the inner-knuckle-print images.The direction of finger is fixed,so the images collected from the same finger only exist tiny translation transformation,which is helpful to improve the precision of the image matching.Moreover,an efficient personal authentication algorithm is put forward.Firstly,the Regions of Interest(ROI) of two fingers are segmented through the line features are detected by Gabor filter.Secondly,a fast image matching algorithm is proposed in the feature matching stage.This algorithm transforms 2-dimension image data to 1-dimension data by projecting the binary image in vertical and horizontal direction,and then aligns ROI images through the comparison of the 1-dimension data.The proposed method is tested on the database which contains 820 samples from 77 different individuals.The equal error rate(EER) reaches 0.61% and one match time consumption is 3.1 ms.The experimental results demonstrate that our proposed approach is rapid and effective.