【導(dǎo)讀】人臉識別是一項當(dāng)下十分熱門的計算機技術(shù)的研究領(lǐng)域，該項技術(shù)可。本身的生物特征從而達到區(qū)分生物體個體的目的。人臉識別技術(shù)目前主要用做身份識別。由于視頻監(jiān)控的飛速普及，使這項應(yīng)。術(shù)，以求能在遠距離之下快速識別人員身份，從而實現(xiàn)智能預(yù)警的功能。背景、發(fā)型等干擾因素，從而定位出人臉區(qū)域。Keywords：Matlab；Grayimage；edgesegmentation；faceregion

　　

【正文】 息電子技術(shù)學(xué)院 29 for i=m:1:1, for j=1:n, if (A(i,j) ~= 0) down = i。 break。 end。 end。 if (down ~= 1) break。 end。 end。 （ 3） 模板匹配 function [ccorr, mfit, RectCoord] = mobanpipei(mult, frontalmodel,ly,wx,cx, cy, angle) frontalmodel=rgb2gray(frontalmodel)。 model_rot = imresize(frontalmodel,[ly wx],39。bilinear39。)。 % 調(diào)整模板大小 model_rot = imrotate(model_rot,angle,39。bilinear39。)。 % 旋轉(zhuǎn)模板 [l,r,u,d] = bianjie(model_rot)。 % 求邊界坐標 bwmodel_rot=imcrop(model_rot,[l u (rl) (du)])。 % 選擇模板人臉區(qū)域 [modx,mody] =center(bwmodel_rot)。 % 求質(zhì)心 [morig, norig] = size(bwmodel_rot)。 % 產(chǎn)生一個覆蓋了人臉模板的灰度圖像 mfit = zeros(size(mult))。 mfitbw = zeros(size(mult))。 [limy, limx] = size(mfit)。 % 計算原圖像中人臉模板的坐標 startx = cxmodx。 starty = cymody。 endx = startx + norig1。 佳木斯大學(xué)學(xué)士學(xué)位論文 佳木斯大學(xué)信息電子技術(shù)學(xué)院 30 endy = starty + morig1。 startx = checklimit(startx,limx)。 starty = checklimit(starty,limy)。 endx = checklimit(endx,limx)。 endy = checklimit(endy,limy)。 for i=starty:endy, for j=startx:endx, mfit(i,j) = model_rot(istarty+1,jstartx+1)。 end。 end。 ccorr = corr2(mfit,mult) % 計算相關(guān)度 [l,r,u,d] = bianjie(bwmodel_rot)。 sx = startx+l。 sy = starty+u。 RectCoord = [sx sy (r1) (du)]。 % 產(chǎn)生矩形坐標 佳木斯大學(xué)學(xué)士學(xué)位論文 佳木斯大學(xué)信息電子技術(shù)學(xué)院 31 附錄 2 外文參考文獻及翻譯 Research development of face recognition at home and abroad Face recognition is one of the most outstanding ability in human visual. Its research involves pattern recognition, image processing, biology, psychology, cognitive science, with other biometric identifica method based on human perception and puter interaction field has a close contact. As early as the sixties and seventies face recognition caused the researchers39。 interest. In the 1960 s, Bledsoe proposed face recognition system of semiautomatic mode and feature extraction method. In the 70 s, the developed countries such as America, Britain began to attach great importance to the research of face recognition and making progress. In 1972, Harmon with interaction is carried out on the face recognition methods in theory and practice. That same year, Sakai face image automatic recognition system is designed. In the early 80 s t. Minami developed better than that of Sakai face image automatic identification system. But early face recognition typically require some prior knowledge of the people, can39。t get rid of the intervention of people. Into the niy s, as a result of the parties face the pressing needs of the face recognition system, the research of face recognition is very popular. Face recognition method is a major breakthrough, entered the stage of the real machine automatic identification such as Karhunen Loeve transform or a new neural work technology. Face recognition research obtained the unprecedented attention, the number of papers published on face recognition and so on increased dramatically, from 1990 to 1990 alone, between SCI and EI can be retrieved as many as thousands of articles, related literature about face recognition during the period of this review is also visible. Abroad, there are many schools in facial recognition technology research, research field is very wide. These studies are 佳木斯大學(xué)學(xué)士學(xué)位論文 佳木斯大學(xué)信息電子技術(shù)學(xué)院 32 military, police and big pany attaches great importance to and support the domestic some wellknown colleges and universities are engaged in the research of face recognition. Face recognition is a frontier topic in the field of pattern recognition, but the current face recognition was still in the stage of research topic, it is not active topic in the field of practical application. While humans can tell a person had no difficulty (in) the human face, but the use of puters for fully automatic face recognition has many difficulties, displays in: the face is a rigid body, the face changes。 Face along with the age growth and change。 The adornment such as hair, glasses of face cause shade。 Face images by affected by illumination and imaging Angle, imaging distance. The difficulty of face recognition is images include a large amount of data, input pixel may be hundreds of thousands, each pixel contains different grayscale, as a result of the plexity of the calculation will increase. Existing recognition methods, by extract features from face image information, to retrieve database method is fast, and the topological properties of map matching is used to determine the matching degree of the method is relatively fast. 佳木斯大學(xué)學(xué)士學(xué)位論文 佳木斯大學(xué)信息電子技術(shù)學(xué)院 33 中文翻譯 國內(nèi)外人臉識別的研究發(fā)展 人臉識別是人類視覺最杰出的能力之一。它的研究涉及模式識別、圖像處理、生物學(xué)、心理學(xué)、認知科學(xué)， 與基于其它生物特征的身份鑒別方法以及計算機人機感知交互領(lǐng)域都有密切聯(lián)系。人臉識別早在六七十年代就引起了研究者的強烈興趣。 20 世紀 60 年代， Bledsoe 提出了人臉識別的半自動系統(tǒng)模式與特征提取方法。 70 年代，美、英等發(fā)達國家開始重視人臉識別的研究工作并取得進展。1972 年， Harmon 用交互人臉識別方法在理論上與實踐上進行了詳細的論述。同年， Sakai 設(shè)計了人臉圖像自動識別系統(tǒng)。 80 年代初 T. Minami 研究出了優(yōu)于Sakai 的人臉圖像自動識別系統(tǒng)。但早期的人臉識別一般都需要人的某些先驗知識，無法擺脫人的 干預(yù)。進入九十年代，由于各方面對人臉識別系統(tǒng)的迫切需求，人臉識別的研究變的非常熱門。人臉識別的方法有了重大突破，進入了真正的機器自動識別階段，如 KarhunenLo232。ve 變換等或新的神經(jīng)網(wǎng)絡(luò)技術(shù)。人臉識別研究得到了前所未有的重視，國際上發(fā)表有關(guān)人臉識別等方面的論文數(shù)量大幅度增加，僅從 1990 年到 2020 年之間， SCI 及 EI 可檢索到的相關(guān)文獻多達數(shù)千篇，這期間關(guān)于人臉識別的綜述也屢屢可見。國外有許多學(xué)校在研究人臉識別技術(shù)，研究涉及的領(lǐng)域很廣。這些研究受到軍方、警方及大公司的高度重視和資助，國內(nèi)的一些知名院校 也開始從事人臉識別的研究。 人臉識別是當(dāng)前模式識別領(lǐng)域的一個前沿課題，但目前人臉識別尚處于研究課題階段，尚不是實用化領(lǐng)域的活躍課題。雖然人類可以毫不困難地由人臉辨別一個人，但利用計算機進行完全自動的人臉識別存在許多困難，其表現(xiàn)在：人臉是非剛體，存在表情變化；人臉隨年齡增長而變化；發(fā)型、眼鏡等裝飾對人臉造成遮擋；人臉所成圖像受光照、成像角度、成像距離等影響。人臉識別的困難還在于圖像包括大量的數(shù)據(jù)，輸入的像素可能成百上千，每個像素都含有各自不同的灰度級，由此帶來的計算的復(fù)雜度將會增加。現(xiàn)有的識別方法中 ，通過從人臉圖像中提取出特征信息，來對數(shù)據(jù)庫進行檢索的方法速度快，而利用拓撲屬性圖匹配來確定匹配度的方法則相對較快。