基于相干增强扩散算法的指纹增强研究

发布时间：2018-04-13 11:47

本文选题：指纹图像增强 + 相干增强扩散　；参考：《太原理工大学》2017年硕士论文

【摘要】：在当前所有生物识别技术中,自动指纹识别技术是其中使用最为普遍的技术之一。它大致包括三个阶段,即指纹图像预处理(指纹图像分割、增强、细化)、特征提取、特征匹配,而通常由于手指外表情况造成的,手指太湿、具有疤痕、过干、手指褪皮或者采集仪器本身的缘故(比如仪器上有污渍)使得采集到的指纹伴有各种不同的噪声,因此需要增强获取到的指纹图像以便于后续步骤的处理。论文研究的重点是指纹增强方法,尤其是使用相干增强扩散的方法来增强指纹图像。论文首先介绍了指纹识别技术,并简要说明了指纹图像增强的重要性,其次总结常用指纹图像增强的算法,如Gabor滤波,短时傅里叶变换滤波等算法。再次,使用平稳小波变换和冲击滤波分别与相干增强扩散相结合的方式增强指纹图像。最后,对当前的工作进行总结以及讨论了研究中还欠缺的问题,并说明今后的研究方向。主要的研究内容有:一、针对相干增强扩散滤波增强的指纹图像会出现边缘模糊以及谷线与脊线间对比度较低的现象,提出使用冲击滤波和相干增强扩散的加权模型方法既能保持相干增强扩散的优势,又能锐化指纹脊线边缘以及增强指纹脊线和谷线的对比度。二、针对相干增强扩散滤波增强的指纹图像会出现边缘模糊以及细节点融合的现象,提出使用在平稳小波域中使用相干增强扩散的方法来增强指纹图像,既能保持相干增强扩散的优势,又能使增强后的指纹图与原图像保持较小的差异而减弱指纹细节点的漂移。
[Abstract]:Automatic fingerprint identification is one of the most popular biometric techniques.It consists of three stages: fingerprint image preprocessing (fingerprint image segmentation, enhancement, thinning, feature extraction, feature matching, and usually due to the appearance of the finger, the finger is too wet, with scars, too dry,Finger peeling or the instrument itself (such as stains on the instrument) make the collected fingerprint accompanied by a variety of noises, so it is necessary to enhance the acquired fingerprint image to facilitate the subsequent processing.The emphasis of this paper is on fingerprint enhancement, especially by using coherent enhancement diffusion to enhance fingerprint image.Firstly, this paper introduces the fingerprint recognition technology, and briefly explains the importance of fingerprint image enhancement, then summarizes the common fingerprint image enhancement algorithms, such as Gabor filtering, short-time Fourier transform filtering and other algorithms.Thirdly, stationary wavelet transform and impulse filter are combined with coherent enhancement diffusion to enhance fingerprint image.Finally, the paper summarizes the current work and discusses the problems which are still lacking in the research, and explains the research direction in the future.The main research contents are as follows: first, the fingerprint image enhanced by coherent enhancement diffusion filter will appear edge blur and low contrast between valley line and ridge line.A weighted model of impulse filtering and coherent enhanced diffusion is proposed, which can not only maintain the advantage of coherent enhanced diffusion, but also sharpen the edge of fingerprint ridge and enhance the contrast between ridge and valley of fingerprint.Secondly, aiming at the phenomenon of edge blur and minutiae fusion in fingerprint image enhanced by coherent enhancement diffusion filter, a method of using coherent enhancement diffusion in stationary wavelet domain is proposed to enhance fingerprint image.It can not only keep the advantage of coherent enhancement diffusion, but also keep small difference between the enhanced fingerprint and the original image, and weaken the drift of fingerprint details.
【学位授予单位】：太原理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】