上海交通大学学报(自然版)

• 自动化技术、计算机技术 • 上一篇    下一篇

基于频域分析的联指图前后背景分离算法

李焱淼1,张永良2,黄亚平3,肖刚2,林璟2
  

  1. (1. 大连交通大学 理学院, 辽宁 大连 116028; 2.浙江工业大学 计算机学院, 杭州 310023;3.浙江工业大学 教育科学与技术学院, 杭州 310032)
  • 收稿日期:2009-12-01 修回日期:1900-01-01 出版日期:2010-08-31 发布日期:2010-08-31

Background Removing of Slap Fingerprint Images Based on Frequency Domain Analysis

LI Yanmiao1,ZHANG Yongliang2,HUANG Yaping3,XIAO Gang2,LIN Jing2
  

  1. (1.School of Science, Dalian Jiaotong University, Dalian 116028, China;
    2.School of Computer Science and Technology, Zhejiang University of Technology, Hangzhou 310023, China;3.School of Education Science and Technology, Zhejiang University of Technology, Hangzhou 310032, China)
  • Received:2009-12-01 Revised:1900-01-01 Online:2010-08-31 Published:2010-08-31

摘要: 为了快速又精确地分离四联指图的前景和背景区域,提出了一种基于频域分析的联指图前后背景分离算法.该算法主要包括:① 基于局部3阶累积量的前后背景粗分离,有效地降低了噪声对方差的影响,并对低对比度区域能够实现有效快速分离;② 结合全局信息的局部最佳直方图阈值分割法的二值化增强算法,实现以局部区域为基础的全局二值化;③ 基于频谱分析和2条判别规则的非指纹区域去除,有效地去除诸如印刷字符之类噪声所在的干扰区域;④ 基于中心最近邻法则的区域合并,使得那些由于质量较差而被分解为若干个不连通的区域标记为同一区域.实验结果表明,该算法的几个步骤是有效的,且具有速度快和精确度高的特点.

关键词: 指纹识别, 10指指纹库, 联指图, 频域分析, 3阶累积量

Abstract: To separate the background from the foreground of slap fingerprint images quickly and accurately, an efficient algorithm based on frequency domain analysis was presented:① the coarse separation of foreground from the background based on threeorder cumulant; ② the binarization enhancement based on optimal threshold segmentation of gray histogram distribution; ③ the removal of nonfingerprint texture region based on frequency domain analysis and two decision rules; ④ the region combination based on the center nearest neighbor rule. The experimental results show that the algorithm, which is proven effective by the given examples, has a faster processing speed and superior precision.

中图分类号: