基于卡口監(jiān)控視頻的人臉特征點(diǎn)定位關(guān)鍵技術(shù)研究
發(fā)布時(shí)間：2018-11-15 07:56
【摘要】：人臉識別技術(shù)是當(dāng)前計(jì)算機(jī)視覺、機(jī)器學(xué)習(xí)等領(lǐng)域的研究熱點(diǎn),在安防、信息安全等領(lǐng)域有十分重要的應(yīng)用背景。其中,人臉特征點(diǎn)定位算法作為人臉識別與驗(yàn)證的關(guān)鍵步驟,與人臉識別準(zhǔn)確率息息相關(guān)。為此,本文以火車站、公交等監(jiān)控卡口為研究對象,針對人臉視角、分辨率變化等特點(diǎn)開展了特征點(diǎn)定位方法的研究。主要工作包括:(1)建立了卡口監(jiān)控人臉數(shù)據(jù)庫。火車站、機(jī)場等監(jiān)控卡口人臉識別是安防領(lǐng)域的難點(diǎn),其圖像通常具有俯視角、分辨率低、人臉姿態(tài)多變、光照條件復(fù)雜等特點(diǎn),導(dǎo)致現(xiàn)有的主流算法難以高效地定位卡口監(jiān)控人臉面部特征點(diǎn)。目前的公開人臉數(shù)據(jù)庫一般是通過爬蟲在互聯(lián)網(wǎng)上采集,與卡口監(jiān)控系統(tǒng)中人臉圖像特點(diǎn)差異較大。為此,本文通過處理大量的火車站和快速公交站監(jiān)控視頻數(shù)據(jù),利用半自動(dòng)的方法對卡口監(jiān)控人臉圖像進(jìn)行檢測、標(biāo)定、篩選處理,最終得到了卡口監(jiān)控人臉數(shù)據(jù)庫。該數(shù)據(jù)庫包括火車站卡口監(jiān)控人臉圖像6647張,快速公交車站卡口監(jiān)控中人臉圖像1287張。本文中涉及到的人臉特征點(diǎn)定位實(shí)驗(yàn)大多利用該卡口監(jiān)控人臉數(shù)據(jù)庫中數(shù)據(jù)進(jìn)行訓(xùn)練、測試和評估。(2)針對局部二值特征算法(Local Binary Features,LBF)的在卡口監(jiān)控中人臉特征點(diǎn)定位存在的問題,提出了基于LBF增量學(xué)習(xí)的卡口人臉特征點(diǎn)定位算法。該方法的主要的思想是在LBF算法級聯(lián)回歸訓(xùn)練的最后一級,利用增量學(xué)習(xí)的方法,向已經(jīng)回歸得到的模型中導(dǎo)入一部分卡口監(jiān)控人臉數(shù)據(jù)庫中的人臉圖像,對新數(shù)據(jù)回歸得到新的形狀增量,從而對現(xiàn)有模型進(jìn)行修正以達(dá)到預(yù)期效果。本文所選取的新加入人臉圖像的數(shù)量是原訓(xùn)練集人臉圖像數(shù)量的十分之一。實(shí)驗(yàn)結(jié)果表明,該方法在卡口監(jiān)控人臉圖像特征點(diǎn)定位上更優(yōu)于時(shí)下主流的監(jiān)督梯度下降法(Supervised Descent Method,SDM)、回歸樹集合算法(Ensemble of Regression Trees,ERT)和LBF算法。(3)針對卡口監(jiān)控人臉的俯視視角、分辨率低和運(yùn)動(dòng)模糊等特點(diǎn),提出了基于權(quán)重自學(xué)習(xí)的多任務(wù)級聯(lián)卷積神經(jīng)網(wǎng)絡(luò)(Multi-task Cascaded Convolutional Networks,MTCNN)算法。該算法首先調(diào)整MTCNN中多任務(wù)的權(quán)重分布,使MTCNN的網(wǎng)絡(luò)結(jié)構(gòu)側(cè)重解決人臉特征點(diǎn)定位問題。然后在MTCNN網(wǎng)絡(luò)結(jié)構(gòu)中加入權(quán)重自學(xué)習(xí)模塊,使其能自動(dòng)學(xué)習(xí)得到多任務(wù)協(xié)調(diào)計(jì)算的最佳權(quán)重分布,從而進(jìn)一步提高對卡口監(jiān)控中人臉圖像特征點(diǎn)定位的精度。實(shí)驗(yàn)結(jié)果表明,該方法在卡口監(jiān)控中人臉圖像特征點(diǎn)定位的準(zhǔn)確率高于SDM、ERT、LBF、MTCNN和基于LBF增量學(xué)習(xí)卡口人臉特征點(diǎn)定位算法。最后對本文工作進(jìn)行了總結(jié),并對本文后續(xù)工作進(jìn)行了展望。
[Abstract]:Face recognition is a hot topic in the field of computer vision and machine learning. It has a very important application background in the field of security and information security. As a key step of face recognition and verification, face feature location algorithm is closely related to the accuracy of face recognition. For this reason, this paper takes the monitoring bayonet such as railway station and bus as the research object, and carries out the research on the feature point location method according to the features of the human face angle of view and the change of the resolution. The main work is as follows: (1) the face database of bayonet monitoring is established. Face recognition of railway station, airport and other monitoring bayonets is a difficult problem in the field of security. The image is usually characterized by low resolution, variable face pose, complex illumination conditions and so on. As a result, the existing mainstream algorithms are difficult to locate the facial feature points efficiently. At present, the open face database is generally collected on the Internet by crawlers, which is different from the features of face images in the bayonet monitoring system. Therefore, through processing a large number of monitoring video data of railway station and bus rapid transit station, this paper uses semi-automatic method to detect, calibrate, filter and process the face image of the bayonet monitoring. Finally, the face database of the bayonet monitoring is obtained. The database includes 6647 face images from railway station bayonets and 1287 face images from bus rapid transit stations. Most of the experiments of facial feature point localization in this paper use the bayonet to monitor the data in face database for training, testing and evaluation. (2) the local binary feature algorithm (Local Binary Features,. In this paper, the problem of face feature point location based on LBF) is discussed. A face feature point location algorithm based on LBF incremental learning is proposed. The main idea of this method is that at the last level of cascade regression training of LBF algorithm, the incremental learning method is used to import a part of the face image in the face database to the model that has been regressed. New shape increments are obtained from the new data regression, and the existing models are modified to achieve the desired results. The number of newly added face images selected in this paper is 1/10 of the original training set. The experimental results show that the proposed method is better than the mainstream supervised gradient descent method (Supervised Descent Method,SDM) and the regression tree set algorithm (Ensemble of Regression Trees,) in the feature point location of face images monitored by the bayonet. ERT) and LBF algorithm. (3) in view of the features of top-down view, low resolution and motion blur, a multi-task concatenated convolution neural network (Multi-task Cascaded Convolutional Networks,MTCNN) algorithm based on weight self-learning is proposed. The algorithm firstly adjusts the weight distribution of multi-task in MTCNN to make the network structure of MTCNN focus on the problem of facial feature point location. Then the weight self-learning module is added to the MTCNN network structure so that it can automatically learn the optimal weight distribution of multi-task coordination calculation and further improve the accuracy of facial image feature point location in the bayonet monitoring. The experimental results show that the accuracy of this method in facial image feature point location is higher than that in SDM,ERT,LBF,MTCNN and LBF incremental learning. Finally, the work of this paper is summarized, and the future work of this paper is prospected.
【學(xué)位授予單位】：集美大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2017
【分類號】：TP391.41
本文編號：2332661
[Abstract]:Face recognition is a hot topic in the field of computer vision and machine learning. It has a very important application background in the field of security and information security. As a key step of face recognition and verification, face feature location algorithm is closely related to the accuracy of face recognition. For this reason, this paper takes the monitoring bayonet such as railway station and bus as the research object, and carries out the research on the feature point location method according to the features of the human face angle of view and the change of the resolution. The main work is as follows: (1) the face database of bayonet monitoring is established. Face recognition of railway station, airport and other monitoring bayonets is a difficult problem in the field of security. The image is usually characterized by low resolution, variable face pose, complex illumination conditions and so on. As a result, the existing mainstream algorithms are difficult to locate the facial feature points efficiently. At present, the open face database is generally collected on the Internet by crawlers, which is different from the features of face images in the bayonet monitoring system. Therefore, through processing a large number of monitoring video data of railway station and bus rapid transit station, this paper uses semi-automatic method to detect, calibrate, filter and process the face image of the bayonet monitoring. Finally, the face database of the bayonet monitoring is obtained. The database includes 6647 face images from railway station bayonets and 1287 face images from bus rapid transit stations. Most of the experiments of facial feature point localization in this paper use the bayonet to monitor the data in face database for training, testing and evaluation. (2) the local binary feature algorithm (Local Binary Features,. In this paper, the problem of face feature point location based on LBF) is discussed. A face feature point location algorithm based on LBF incremental learning is proposed. The main idea of this method is that at the last level of cascade regression training of LBF algorithm, the incremental learning method is used to import a part of the face image in the face database to the model that has been regressed. New shape increments are obtained from the new data regression, and the existing models are modified to achieve the desired results. The number of newly added face images selected in this paper is 1/10 of the original training set. The experimental results show that the proposed method is better than the mainstream supervised gradient descent method (Supervised Descent Method,SDM) and the regression tree set algorithm (Ensemble of Regression Trees,) in the feature point location of face images monitored by the bayonet. ERT) and LBF algorithm. (3) in view of the features of top-down view, low resolution and motion blur, a multi-task concatenated convolution neural network (Multi-task Cascaded Convolutional Networks,MTCNN) algorithm based on weight self-learning is proposed. The algorithm firstly adjusts the weight distribution of multi-task in MTCNN to make the network structure of MTCNN focus on the problem of facial feature point location. Then the weight self-learning module is added to the MTCNN network structure so that it can automatically learn the optimal weight distribution of multi-task coordination calculation and further improve the accuracy of facial image feature point location in the bayonet monitoring. The experimental results show that the accuracy of this method in facial image feature point location is higher than that in SDM,ERT,LBF,MTCNN and LBF incremental learning. Finally, the work of this paper is summarized, and the future work of this paper is prospected.
【學(xué)位授予單位】：集美大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2017
【分類號】：TP391.41
【參考文獻(xiàn)】
相關(guān)期刊論文 前2條
1 劉仁明;毛建旭;歷艷琨;;基于隨機(jī)森林回歸的人臉特征點(diǎn)定位[J];電子測量與儀器學(xué)報(bào);2016年05期
2 宋剛,艾海舟,徐光yP;紋理約束下的人臉特征點(diǎn)跟蹤[J];軟件學(xué)報(bào);2004年11期
相關(guān)博士學(xué)位論文 前4條
1 崔瀅;基于統(tǒng)計(jì)學(xué)習(xí)的人臉特征點(diǎn)定位算法研究[D];南京理工大學(xué);2015年
2 牛志恒;面部特征點(diǎn)定位方法研究[D];哈爾濱工業(yè)大學(xué);2009年
3 杜春華;人臉特征點(diǎn)定位及識別的研究[D];上海交通大學(xué);2008年
4 山世光;人臉識別中若干關(guān)鍵問題的研究[D];中國科學(xué)院研究生院（計(jì)算技術(shù)研究所）;2004年
相關(guān)碩士學(xué)位論文 前1條
1 周帆;人臉特征點(diǎn)跟蹤方法的研究[D];西安電子科技大學(xué);2015年
，本文編號：2332661
本文鏈接：http://sikaile.net/kejilunwen/ruanjiangongchenglunwen/2332661.html
最近更新
教材專著
