摘要
针对干扰情况下指纹图像边缘提取准确性较差的问题,基于指纹特殊的纹理特征,设计了一种基于二维经验模式分解(BEMD)的指纹边缘检测算法。首先通过BEMD将指纹图像分解成具有不同特征尺度的、从高频到低频分布的固有模态函数(IMF)分量和一个残余分量,鉴于高频分量既能有效抑制非对称扰动干扰,又能较好地保留指纹图像的细节特征,接下来取高频IMF分量作为处理对象,通过对获得的IMF分量进行二值化、细化处理,得到指纹边缘检测结果。与传统的边缘检测算法相比,获得的指纹纹线清晰度得到了有效改善,较好保留了指纹的细节特征。
In the view of problem of poor accuracy of fingerprint image edge extraction under condition of interference,based on special form of fingerprint texture,it is improved by using a kind of method based on the bidimensional empirical mode decomposition（ BEMD） to extract fingerprint edge. BEMD make fingerprint image decompose into intrinsic mode function（ IMF） components and a residual component,since high frequency component can effectively restrain interference asymmetric disturbances and reserve the detail of the fingerprint image characteristics,so it can be acted as processing object and obtained IMF component binarization processing and refine treatment,obtain the fingerprint edge detection results. Compared with traditional edge detection algorithm,fingerprint ridge clarity is effectively improved and detail features are kept very well.
出处
《传感器与微系统》
CSCD
2016年第10期127-130,134,共5页
Transducer and Microsystem Technologies
基金
国家自然科学基金资助项目(61301091)
关键词
二维经验模式分解
高频
固有模态函数分量
边缘检测
bidimensional empirical mode decomposition（BEMD）
high frequency
intrinsic mode function（IMF） component
edge detection
