【作者】 杨德英；

【导师】 卢朝阳；

【作者基本信息】 西安电子科技大学 ， 通信与信息系统， 2005， 硕士

【摘要】 指纹作为一种独特的身份特征已经在很长一段时间内得到了广泛应用。政府机构(如公安系统)及许多需要高度安全的场所不断收集大量的指纹图像以用于传输、存储和自动分析识别。本文所述的部分是指纹图像的预处理和增强方面的一些研究工作。 去除指纹图像中的背景以及不可恢复区域，可以缩短图像预处理时间，提高特征提取的准确率。本文提出了一种基于二维最大类间方差的指纹图像分割算法。首先利用指纹方向一致性特征将指纹图像转换成具有较大灰度均匀区域的灰度图像。然后运用二维最大类间方差法对变换后的图像进行分割，提取出了指纹纹线区域。最后利用格雷厄姆凸壳算法对提取出的指纹前景轮廓线进行平滑。 基于指纹图像的频谱特性，我们采用了具有良好的方向和频率选择特性的Gabor滤波器对指纹图像进行增强处理。采用最小均值平方估计法估计局部区域中纹线的主导方向：灰度投影分析法估计局部区域的纹线频率。根据求得的纹线方向和纹线频率设置Gabor滤波器模板，对指纹图像进行滤波处理。另外，在奇异点区域，我们采用了一个方向无关性的滤波器来增强这个特殊的区域。 在论文的最后，我们总结了全文，指出了目前研究工作中需要进一步完善的地方以及今后工作的研究方向。更多还原

【Abstract】 A fingerprint image enhancement system is described in this paper.Fingerprint is unique identifiers of individuals and has been used for a long time. Government agencies (such as the police) and many kinds of departments needing high security have to collect increasing amount of fingerprints for transmission, storage, and automated analysis. We need some preprocessing and some enhancement technique to improve the recognition of fingerprint image.Get ride of the background and unrecoverable region can not only reduce the time of subsequent processing, but also improve the reliability of feature extraction considerably. A fingerprint image segmentation method based on 2D maximum between-cluster variance is suggested in this paper. First, we use coherence of orientation to turn fingerprint image into image with greater even gray areas. Then we successfully achieve the fingerprint region with image segmentation method based on 2D maximum between-cluster variance. Last, we use Graham convex hull algorithm to smooth the contour of fingerprint region.Based on character of fingerprint, Gabor filters with frequency-selective and orientation-selective properties are used to enhance fingerprint image. We estimate the orient field by a least mean square algorithm and estimate local ridge frequency by projection method. According with estimated orient field and ridge frequency, we use Gabor filters to enhance fingerprint image. In the region of singular point, we adopt a different filter to enhance.In the last part of this dissertation, summarize the work of this dissertation, analyze the improvements need to be done and give the directions of future work.更多还原