中文摘要：

自动指纹识别系统（AFIS）的性能严重依赖于输入指纹的质量,因此有效指纹增强算法对该系统具有重要意义。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,以及增强指纹脊线与谷线的对比度。增强的主要过程为,建立一个相干增强扩散和冲击滤波的加权组合模型,加权函数是以指纹梯度为自变量,使得在扩散过程中在图像边缘处以冲击滤波为主,在脊线与谷线内部则以相干增强扩散为主。实验表明,使用这种方法,能得到更加清晰的指纹图像,便于之后的二值化与细化过程处理,使得自动指纹识别系统具有更好的性能。

英文摘要：

The performance of automatic fingerprint identification system is heavily determined by the quality of the input image, thus an effective method to enhance the fingerprint image is essential in such a system. Since the method of fingerprint image enhancement of coherence-enhancing diffusion has the disadvantages that the enhanced image will appear blurring edges and low contrast between ridges and valleys of fingerprint images, a new method combining coherence-enhancing diffusion and shock filter is put forward which can not only maintain an advantage of coherence-enhancing diffusion, but sharpen edges and enhance the contrast between ridges and valleys. Thus no-vel approach comes out. First, a weighting model of combining coherence-enhancing diffusion with shock filter is bulit, which emphasizes particularly on shock filter at edges while on coherence-enhancing diffusion at the other part, and gradient is taken as automatic variable of weighting function. Experimental results show that the proposed combining approach can obtain a more clear fingerprint image which is convenient to the process of binarization and thinning and finally achieve a better performance of the system.