- 相關(guān)推薦
指紋識別算法研究
畢業(yè)論文
摘 要
指紋具有唯1性和穩(wěn)定性，因此被人們用來當(dāng)作鑒別個人身份的主要依據(jù)。自動指紋識別系統(tǒng)是基于計(jì)算機(jī)來進(jìn)行指紋識別的技術(shù)，具有方便、高效、安全、可靠等優(yōu)點(diǎn)，在金融安全、數(shù)據(jù)加密、電子商務(wù)等各個領(lǐng)域都得到了廣泛的應(yīng)用，并將在我們的生產(chǎn)和生活中發(fā)揮越來越重要的作用。
本文的內(nèi)容正是關(guān)于自動指紋識別系統(tǒng)的研究，按照設(shè)計(jì)過程，指紋識別主要包括3個大部分:指紋圖像的預(yù)處理、特征提取以及匹配。
指紋圖像的預(yù)處理又可以分為灰度圖濾波去噪、2值化、2值化圖像去噪、細(xì)化和細(xì)化后去噪5個部分。本文先基于指紋的方向圖設(shè)計(jì)出方向?yàn)V波器對原圖像進(jìn)行濾波去噪，然后使用局部平滑閾值自適應(yīng)2值化算法，將灰度圖像進(jìn)行2值化，并采用快速傅氏變換對所得到的2值化圖像進(jìn)行去噪處理。接下來使用細(xì)化模板對2值化圖像進(jìn)行細(xì)化，并針對細(xì)化圖中各種噪聲的拓?fù)浣Y(jié)構(gòu)將它們11濾除。
指紋圖像的特征提取主要是提取指紋的細(xì)節(jié)特征及其位置。本文先采用脊線跟蹤法將指紋圖中的細(xì)節(jié)特征全部找出來，再對每個細(xì)節(jié)特征進(jìn)行驗(yàn)證，盡量去除偽特征點(diǎn)。然后采用求Poincare Index值的方法確定指紋的中心點(diǎn)，并作為參照點(diǎn)來確定每個特征點(diǎn)相對參照點(diǎn)的位置。
指紋圖像的匹配過程包括了圖像校準(zhǔn)和細(xì)節(jié)匹配兩個部分。首先，找到輸入圖像和模板圖像的參照點(diǎn)對，然后將兩幅圖像中的細(xì)節(jié)特征點(diǎn)相對于各自的參照點(diǎn)轉(zhuǎn)化為極坐標(biāo)形式，最后進(jìn)行比對，確定兩幅圖像是否來自于同1手指。
關(guān)鍵詞：預(yù)處理；特征提取；匹配；2值化；細(xì)化；細(xì)節(jié)特征
Abstract
Due to the uniqueness and persistence, fingerprint is used as main basis of personal identity. Automated fingerprint identification system is a technology of fingerprint identification by computer, which is of convenience, high efficiency, security and reliability. It has been applied in many fields such as financial security, data encryption, electronical business and some, and will play a more and more important role in our life.
The paper is about the study of automated fingerprint identification system. According to the process of the design, the paper can be devided into three components: pre-processing, feature extraction, matching of fingerprint images.
Fingerprint image pre-processing has five parts: filtration in gray-scale image, binarization, filtration in binary image, thinning and filtration in thinning image. In the paper, we firstly design orientation filters based on directional image of fingerprint and employ them to denoise gray-scale image. Then, we binarize the gray-scale image with local self-adaptive binarization smoothness algorithm andeliminate the noises from the binary image with fast Fourier transform algorithm.Afterwards, by using thinning templates, we get the skeleton fingerprint imagefrom the binary image. After thinning, we get rid of the noises from the acquired skeleton image according to their configuration.
Fingerprint image feature extraction mainly extracts the minutiae and their positions. Firstly, this paper presents an algorithm based on ridge following to extract all minutiae from the pre-processed image. Secondly, we validate these minutiae and eliminate pseudo ones. Then, by computing the value of Poincare Index, we can find the core of the fingerprint. Finally, we can fix on the relative positions of the minutiae according to the core.
Fingerprint image matching has two steps: image adjustment and minutiae matching. First of all, We select a referrence point pair of the input image and the template image. And then we transform the minutiae positions into polar coordinates. Finally, we match the input image with the template one to judge whether these two images are captured from the same finger.
Keywords: pre-processing; feature extraction; matching; binarization; thinning; minutiae
【指紋識別算法研究】相關(guān)文章：
計(jì)數(shù)查找算法的研究11-22
LDPC碼譯碼算法研究03-07
紅外圖像增強(qiáng)算法研究03-07
FFT算法的研究與DSP實(shí)現(xiàn)03-07
iLBC語音算法的初步研究03-07
接力切換的基本算法及流程研究03-07