摘要
掌纹特征的提取在掌纹自动识别系统中是一项必不可少的重要环节.首先对人手图像进行预处理,提取出ROI区域,并针对掌纹图像噪声强、对比度低的特点,先根据图像灰度特征运用局部自适应的二值化方法提取出掌纹的主线特征;再针对存在的部分噪声点和断点的情况,运用提出的邻域法跟踪出每条主线附近的点,方便地剔除了噪声点;然后对这些点进行多项式拟合处理,提取出细化的主线,连接了断点.实验表明:新的提取方法能够有效地去除干扰点及连接纹线的断点,且提取效果误差小,接近自然掌纹主线.
The feature extraction of palmprint acts as an important step in automatic palmprint recognition system. In this article, pretreatment is carried out first for the hand picture to obtain the ROI region. As the palm picture is corrupted with strong noise and low contrast, the principal lines are first singled out with self-adaptive local binarization based on the characteristic of the gray＇ image. To tackle some existing noises and breaking points, a neighborhood algorithm is proposed to track the points close to every principal line. In so doing, the noises can thus be eliminated with ease. Next curve-fitting is taken on these points to acquire the thinned principal line and connect the breaking points. The experiment results indicate that these methods can efficiently remove the noise and connect breaking points, and only contain insignificant error and sufficiently well match the natural principal lines appeared in palmprints.
出处
《宁波大学学报（理工版）》
CAS
2008年第4期462-465,共4页
Journal of Ningbo University：Natural Science and Engineering Edition
基金
国家自然科学基金(10571095)
浙江省新苗人才计划项目基金(2007G60G2070040)
关键词
掌纹主线
局部二值化
邻域
principal line of the palmprint
local binarization
neighborhood
