摘要
针对采用分形维数作为特征描述掌纹信息不准确的问题,对差分盒子维进行改进提高特征区分性。此外,由于采用单一的特征不足以描述掌纹纹理,引入Gabor变换,提出一种基于Gabor变换与改进差分盒子维(GIDBC,Gabor improved differential box counting)相结合的掌纹识别算法。通过在PolyU掌纹图像库上实验,与传统高性能算法比较,本算法识别率最高可达到99.78%,表明了本文方法的有效性,同时特征提取与匹配时间为338 ms,满足实时性要求。
In order to solve the problem that traditional fractal dimension as characterization pahnprint information is not accurate,the differential box dimension is improved to enhance the distinguish ability of feature.In addition,the pahnprint feature is not enough to describe the texture of palmprint,hence,Gabor transform is introduced and a novel palmprint recognition algorithm based on Gabor transform and improved differential box dimension（GIDBC） is proposed.By carrying out the related experiments on PolyU palmprint database,compared with those traditional high-performance algorithms,the recognition accuracy can reach 99.78%,which demonstrates the effectiveness of the proposed method.Meanwhile,feature extraction and matching time is 338 ms,which meets the real-time requirement.
出处
《测控技术》
CSCD
2015年第9期38-41,共4页
Measurement & Control Technology
基金
四川省科技计划项目基金(2011GZ0022)
关键词
掌纹识别
分形维数
改进差分盒子维
GABOR变换
palmprint recognition
fractal dimension
improved differential box dimension
Gabor transform
