摘要
指纹图像的分割是指纹识别系统中的重要环节之一,分割算法的优劣直接影响指纹自动识别系统的性能,因此指纹分割技术的研究有着重要的意义。一幅指纹图像,在指纹区灰度变化剧烈,在垂直于指纹局部方向上有着稳定的周期性。现对梯度和指纹周期设定阈值,提出了一种融合梯度和周期两种特征的分割算法。在梯度的计算上改用旋转不变的Sobel算子;在局部周期的计算上,通过对傅里叶变换的分析推理,提出了一种新的周期计算方法。分析、处理结果都表明:该算法抗噪能力强、分割效果好,具有很高的实用价值,能够满足一般指纹识别系统性能要求。
The segmentation of fingerprint images is the key step in automatic fingerprint identification system. The performance of segment algorithm directly affects performance of the automatic fingerprint identification system. So it is significant to research on the technology of segmentation of fingerprint images. The intensity in foreground changes rapidly in fingerprint image, and there is a steady period in the orthogonal direction to the local ridge orientation. A segment algorithm based on fusion of features - gradient and period is proposed. A new Sobel operator is used to compute the gradient; a new method of computing the local period is proposed after analyzing the Fourier transformation. The result of the analysis and experiment shows that the segment algorithm not only provides an good segment result but also is robust to noise, and it holds higher practical value, can satisfy the requirements of the fingerprint identification system.
出处
《信息技术》
2007年第3期32-35,共4页
Information Technology
关键词
指纹识别
局部周期
梯度
特征融合
fingerprint identification
local period
grad
fusion of features
