复杂背景下基于二维OTSU和肤色分割结合的掌纹图像分割方法  

Palmprint image segmentation in complex background based on 2-D OTSU and skin color segmentation

作　　者：薛延学[1] 刘敏 马思欣 帅建坤 雷丹[1] 

机构地区：[1]西安理工大学印刷包装与数字媒体学院,陕西西安710048

出　　处：《西安理工大学学报》2015年第3期353-359,共7页Journal of Xi'an University of Technology

基　　金：西安市科技计划项目(CXY1440-7)

摘　　要：基于智能手机的掌纹识别符合非接触式、低分辨率等生物特征认证的发展趋势,但是由智能手机采集的掌纹图像背景往往比较复杂,导致手掌区域不能正确分割,限制了掌纹识别技术在智能手机上的应用推广。针对上述问题,本文在研究二维大津法(二维OTSU)和基于高斯模型的肤色分割两种方法的基础上,结合两者的优点,将两种方法分别以"串联"和"并联"两种融合方式相结合提出一种新的方法,对使用智能手机采集的复杂背景下掌纹图像分别进行分割。经实验验证,并联融合的方法对背景偏暗的掌纹图像分割正确率较高;串联融合的方法对背景偏亮掌纹图像正确分割率较高,两种方式结合能很好地满足复杂背景下掌纹分割的需求。Palmprint identification system based on smart-phones is in accord with the develop-ment trend of biometric recognition with non-contact,low-resolution and other features.Howev-er,the background of the palmprint image captured by smart phones tends to be complex,cau-sing the palm area fail to be properly segmented by system,which limits the application of palm-print recognition technology for intelligent mobile phone.To solve the above problem,a new ap-proach is proposed in this paper to split the palmprint image captured by smart phones with com-plex background based on the analysis of 2-D OTSU and skin color segmentation based on gaussi-an model and combination of the two methods by the means of“series”and“parallel”respective-ly.The experimental results prove that “parallel”of methods provides higher accuracy rate for splitting the palmprint image with dim background,whereas “series”combination of methods provides higher accuracy rate for splitting the palmprint image with bright background.The com-bination of the two methods can fully meet the needs of splitting the palmprint image with com-plicated background.The new approach achieves the ideal effect under the conditions of complex.

关 键 词：智能手机 复杂背景 手掌区域 掌纹图像分割 二维大津法 肤色分割 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]