一种指纹图像综合增强方法  

A comprehensive enhancement approach of fingerprint images

作　　者：苏铁明[1] 韩兆翠[1] 魏鸿磊[2] 

机构地区：[1]大连理工大学机械工程学院,辽宁大连116024 [2]大连水产学院机械工程学院,辽宁大连116023

出　　处：《大连理工大学学报》2010年第5期686-691,共6页Journal of Dalian University of Technology

基　　金：大连理工大学-中国科学院沈阳自动化研究所联合基金资助项目(2006)

摘　　要：图像去污增强处理应当基于其内容特点和应用需要.指纹图像重要的识别用特征是其纹线方向场及其细节点.根据指纹图像的特点提出并研究了一种指纹增强算法.首先,通过梯度一致性检验去除图像中的背景区和纹理无序区.然后,通过曲线波变换、滤波处理和Retinex处理,增强纹理的曲线形状轮廓并规整化图像灰度差异.最后,经过二值化细化,获得清晰指纹图像.实测的图例表明该算法有效,图像去污增强质量良好.Denoising and enhancements of images should be processed based on their contents and the application focuses.The important features of fingerprint images for identification are texture orientation fields and their minutiae.According to the characteristics of fingerprint,an enhancement approach dedicated for fingerprint identification is presented.First,the background region and the texture chaos regions are removed based on the gradient consistency.Then,a Curvelet-based filtering processing and a Retinex processing are implemented,which are developed aiming at enhancing the curve-like texture contour,removing the noise and normalizing the image intensities.Finally,a binary thinning processing is implemented to get the clear result image.The testing examples show that the enhancement and denoising approach is effective and the qualities of result images are quite good.

关 键 词：指纹 图像增强 曲线波 RETINEX 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]