作　　者： ; ; ;

机构地区： 华南理工大学自动化科学与工程学院

出　　处： 《计算机工程与设计》 2011年第2期653-655,670,共4页

摘　　要： 为了快捷而准确地提取掌纹线,提出了一种基于灰度差统计分析的方法。采用了带有修正因子的直方图均衡化方法使掌纹图像灰度分布均匀化,并应用基于邻域平均灰度的计算方法有效抑制伪纹线的干扰。在预先设置长度阈值的情况下,对图像灰度差值图进行统计分析并设置灰度阈值,进而使用具有方向性的8邻域搜索方法,通过判断对象点的灰度值和连续掌纹线点集的点数,提取出掌纹线的二值图像。最后通过计算掌纹图像间的隶属度评价掌纹线的提取效果。实验结果表明,该方法提取出的掌纹线图像清晰,识别正确率较高,达到94.51%以上。 For fast and accurate extraction ofpalmprint, a new method based on statistical analysis of gray level difference is proposed. A histogram equalization method with correction factor is firstly adopted to homogenize the gray level distribution. The fake lines are decreased effectively by calculations that based on mean value of neighborhood. Secondly with the length threshold configured, a directional eight neighborhood search approach is used to extract the palmprint, while its gray threshold is set by analyzing the gray level difference statistics. The degree of membership among different images is lastly applied to evaluate the effect. The method is experimentally proved to work well with a high accuracy rate that could be higher than 94.51%.

关 键 词： 图像识别 生物认证 掌纹线提取 直方图均衡化 灰度差统计

领　　域： [自动化与计算机技术] [自动化与计算机技术]