CAJ | 학술논문

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。
기우지능수궤적장문식별부합비접촉식、저분변솔등생물특정인증적발전추세，단시유지능수궤채집적장문도상배경왕왕비교복잡，도치수장구역불능정학분할，한제료장문식별기술재지능수궤상적응용추엄。침대상술문제，본문재연구이유대진법（이유 OTSU）화기우고사모형적부색분할량충방법적기출상，결합량자적우점，장량충방법분별이“천련”화“병련”량충융합방식상결합제출일충신적방법，대사용지능수궤채집적복잡배경하장문도상분별진행분할。경실험험증，병련융합적방법대배경편암적장문도상분할정학솔교고；천련융합적방법대배경편량장문도상정학분할솔교고，량충방식결합능흔호지만족복잡배경하장문분할적수구。
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.