摘要
方向图(场) (directionalimage/field)是自动指纹识别中的一项关键技术,为了进一步提高指纹方向图的精确度,提出了指纹连续分布方向图(场)的概念及其算法。该算法是利用直方图滤波技术对指纹点方向图进行滤波处理,得到连续分布方向图。实践证明这种连续分布方向图过渡平滑、自然,既具有很好的连续性、渐变性和抗噪性,又具有极高的精确度,还兼有指纹点方向图的连续过渡性和块方向图的抗噪性优点。实验结果表明,由于连续分布方向图能更精确地反映指纹脊线的结构特征,且能有效地解决指纹图像方向图的连续过渡性、精确性和抗噪性问题,因而为自动指纹识别系统(AFIS)中高精度的指纹图像增强、二值化、全局特征的提取、自动分类、模板匹配、指纹图像编码等许多关键处理步骤奠定了良好的基础。
Fingerprint directional image/field is a key technique of automated fingerprint identification system (AFIS). In order to improve the accuracy of fingerprint directional image further, combining the advantage of the continuity of the point directional image with the advantage of the robustness of the block directional image to the noises in the fingerprint image, a novel concept on the continuously distributed directional image/field (CDDF) and the method to compute it in the fingerprint images are developed, which transits smoothly and exhibits not only good continuity, well gradualness and excellent robustness to the noises, but very high precision, also. Therefore, it can effectively overcome the defect of low robustness to the noises of the point directional image and the weakness of low precision of the block directional image of the fingerprint image. The experimental results prove that the continuously distributed directional image can not only filter the noises in point directional image, but also represent the basic structural feature of fingerprint more precisely. Furthermore, it can solve the key problem of the continuity, the precision and the performance of robustness of fingerprint directional fields effectively, and lays the base for high precision of fingerprint image enhancement, binarization, global feature extraction, automatic classification, template matching and image encoding in AFIS.
出处
《中国图象图形学报（A辑）》
CSCD
北大核心
2005年第3期315-319,F007,共6页
Journal of Image and Graphics
基金
江苏省教育厅自然科学基金项目(院编 2002316)
关键词
自动指纹识别系统
指纹图像
指纹方向图
算法
模板匹配
二值化
自动分类
抗噪性
处理步骤
滤波技术
biometrics, automated fingerprint identification system(AFIS), continuously distributed directional image/field(CDDF), preprocessing of fingerprint image, noise cleaning
