一种针对低质量指纹证据图像的增强算法设计  

Design of an enhancement algorithm for low quality fingerprint evidence images

作　　者：卢超超 Lu Chaochao(Department of Investigation,Shanxi Police College,Taiyuan 030401,China)

机构地区：[1]山西警察学院侦查系,太原030401

出　　处：《现代计算机》2024年第7期17-23,共7页Modern Computer

摘　　要：指纹图像采集过程中存在旋转、平移以及图像采集不全等内在差异,导致每次提取到的细节点数目是不固定的,这种特征形式后期匹配计算复杂度较高。当前基于模板匹配的图像增强算法针对低质量指纹证据图像匹配计算复杂度较高的情况,以固定阈值完成原始图像滤波处理,未考虑单像素细化方向场特征丢失问题。提出基于卡尔曼滤波的低质量指纹证据图像增强方法。根据图像高斯噪声分布函数,选用VisuShrink阈值降噪,构建卡尔曼滤波器,完成指纹图像预处理。将原始图像转化为单像素细化图,提取滤波后指纹图像的奇异点与特征点。划分指纹图像,根据奇异点与方向向量确定指纹的方向场,进行图像增强滤波处理,完成基于卡尔曼滤波的低质量指纹证据图像增强方法设计。构建实验环节,验证该方法应用效果,实验方法表明:该方法使用后,低质量指纹图像变得更加清晰,并且平均错误率最高仅为6.21%,评估分数均值达到了0.85,该方法有效提高了图像的增强效果。During the process of fingerprint image acquisition,there are inherent differences such as rotation,translation,and incomplete image acquisition.The number of fine nodes extracted each time is not fixed,and this feature form has a high computa⁃tional complexity for later matching.The current image enhancement algorithms based on template matching aim to address the high computational complexity of low⁃quality fingerprint evidence image matching,and complete the original image filtering process with a fixed threshold without considering the issue of single pixel thinning direction field feature loss.Propose a low quality finger⁃print evidence image enhancement method based on Kalman filtering.Based on the Gaussian noise distribution function of the im⁃age,VisuShrink threshold denoising is selected to construct a Kalman filter and complete fingerprint image preprocessing.Convert the original image into a single pixel refined image,and extract the singular points and feature points of the filtered fingerprint im⁃age.Divide fingerprint images,determine the direction field of the fingerprint based on singular points and direction vectors,per⁃form image enhancement filtering processing,and complete the design of a low⁃quality fingerprint evidence image enhancement method based on Kalman filtering.Construct an experimental section to verify the application effect of this method.The experimen⁃tal method shows that after using this method,low⁃quality fingerprint images become clearer,and the average error rate is only 6.21%,with an average evaluation score of 0.85.This method effectively improves the image enhancement effect.

关 键 词：卡尔曼滤波 指纹增强 图像增强 方向场 特征点提取 图像预处理 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]