一种掌纹图像感兴趣区域的提取方法

Method for region of interest extraction of the palmprint images

掌纹识别是生物特征识别中的一种,由于其分辨率要求低、设备成本低、用户易接受等优点受到众多研究者的关注。从一幅掌纹图像中,找出感兴趣的目标区域(ROI),使之从图像背景中分离出来,这是掌纹特征提取和进行进一步匹配的关键步骤。论文提出一种新的感兴趣区域的提取方法:首先使用全局阈值对原掌纹图像进行二值化处理,并提取出手掌的边缘,然后使用食指和中指、无名指和小指之间的指缝确定一个斜率,再计算出手掌部分的中心,并据此建立一个新的坐标系,最后以此坐标系原点为中心提取出一个128×128的感兴趣区域。实验结果证实了该方法与相关文献所提方法相比更加有效。

Palmprint recognition,a kind of biometrics,recently receives many researchers’attention because of its low resolution, cheap devices and easily accepted by the users.The key step that finding a region of interesting（ROI） and segmenting from the background of the palmprint image is important for the feature extraction and further matching.This paper presents a new method of extracting the ROI.Firstly,a global threshold is used to binarize the original palmprint image,and the edge of the palm is extracted. Then,a new coordinate system is settled according to the slope making by the two gaps between the first and middle fingers,the ring and little fingers,and the centre of the palm.Finally,a fixed 128×128 subimage is extracted as ROI.Compared with other methods which correlative papers presented,the experimental results demonstrate the effectiveness and accuracy of the proposed method.