结合自适应肤色模型和区域生长的掌纹分割

摘要

为克服移动终端掌纹图像预处理受到的多方面干扰,设计一种“工”字形辅助定位拍摄方法,在此基础上提出一种结合自适应高斯肤色模型和区域生长的掌纹分割方法.选取训练子区域自适应估计高斯肤色模型的参数,设定肤色似然度阈值,得到非完整的分割结果,进行区域生长,分割出准确的手部区域.该方法可有效抵御与手部不连通背景类肤色区域的干扰,克服单一固定闽值分割对手部区域肤色渐变的失效性.实验结果表明,该方法在移动环境下能够达到较高的正确率和认证精度.%The capture technique of I-shaped auxiliary location was developed to overcome the multiple interferences of palmprint preprocessing of mobile terminal.Furthermore,palmprint region was segmented combining adaptive Gaussian skin-color model and region growing.The parameters of Gaussian model were calculated through training sub-image,and the likelihood threshold of skin-color was used to segment non-complete palmprint region.Region growing was used in the non-complete region and complete palmprint region was segmented.The proposed method resists the skin-color-like interference regions in background,which are not connected to the hand.It overcomes the failure of one fixed threshold for gradual change in hand region.Experimental resuits confirm that the proposed method can achieve higher correctness and verification accuracy in mobile environments.