CBCL FACE DATABASE #1(CBCL人脸数据库#1)_图像处理_科研数据集
- 格式:pdf
- 大小:67.68 KB
- 文档页数:4
Labeled Faces in the Wild Home(户外脸部检测数据库)数据摘要:Welcome to Labeled Faces in the Wild, a database of face photographs designed for studying the problem of unconstrained face recognition. The data set contains more than 13,000 images of faces collected from the web. Each face has been labeled with the name of the person pictured. 1680 of the people pictured have two or more distinct photos in the data set. The only constraint on these faces is that they were detected by the Viola-Jones face detector. More details can be found in the technical report below.中文关键词:脸部,识别,户外,名字,标注,英文关键词:face,recognition,wild,name,labeled,数据格式:IMAGE数据用途:studying the problem of unconstrained face recognition数据详细介绍:Labeled Faces in the Wild HomeWelcome to Labeled Faces in the Wild, a database of face photographs designed for studying the problem of unconstrained face recognition. The data set contains more than 13,000 images of faces collected from the web. Each face has been labeled with the name of the person pictured. 1680 of the people pictured have two or more distinct photos in the data set. The only constraint on these faces is that they were detected by the Viola-Jones face detector. More details can be found in the technical report below.Mailing list:If you wish to receive announcements regarding any changes made to the LFW database, please send email to majordomo@ with the message body: "subscribe lfw" on a single line.Explore the database:Alphabetically by first name:[A)[Alf)[Ang)[B)[Bin)[C)[Che)[Col)[D)[Daw)[Don)[E)[Eri)[F)[G)[Goe)[H)[I)[J)[Jav)[Jes)[Joh)[Jos)[K)[Kim)[L)[Lil)[M)[Mark)[Mel)[Mik)[N)[O)[P)[Per)[Q)[R)[Ric)[Rog)[S)[Sha)[Ste)[T)[Tim)[U)[V)[W)[X)[Y)[Z)Alphabetically by first name, only people with more than one image:[A][B][C][D][E][F][G][H][I][J][K][L][M][N][O][P][Q][R][S][T][U][V][W][X][Y][Z] Alphabetically by last name:[A][B][C][D][E][F][G][H][I][J][K][L][M][N][O][P][Q][R][S][T][U][V][W][X][Y][Z]By number of images per person:[1 A-E][1 F-J][1 K-O][1 P-T][1 U-Z][2][3][4][5][6-10][11+]Single page of all names (no thumbnails)Download the database:∙All images as gzipped tar file (173MB, md5sumac79dc88658530a91423ebbba2b07bf3)∙All images aligned with funneling (233MB, md5sum1b42dfed7d15c9b2dd63d5e5840c86ad)∙All images aligned with commercial face alignment software (LFW-a - Taigman, Wolf, Hassner)∙Subset of images - people with name starting with A (14MB) as zip file∙Subset of images - George_W_Bush (individual person with most images) (6.9MB)as zip file∙All names (with number of images for given name) as text file∙README - information on file formats and directory structure Training, Validation, and Testing:View 1:For development purposes, we recommend using the below training/testing split, which was generated randomly and independently of the splits for 10-fold cross validation, to avoid unfairly overfitting to the sets above during development. For instance, these sets may be viewed as a model selection set and a validation set. See the tech report below for more details.Explore the sets: [training][test]Download the sets: pairsDevTrain.txt, pairsDevTest.txt, peopleDevTrain.txt, peopleDevTest.txtView 2:As a benchmark for comparison, we suggest reporting performance as 10-fold cross validation using splits we have randomly generated.For information on the file formats, please refer to the README above.For details on how the sets were created, please refer to the tech report below. Results:Accuracy and ROC curves for various methods available on results page. Information:13233 images5749 people1680 people with two or more imagesErrata:The following is a list of known errors in LFW. Due to the small number of such errors, the database will be left as is (without corrections) to avoid confusion.It is important that users of the database provide their algorithms with the database as is, i.e. without correcting the errors below, since previous results published for the database did not have the advantage of correcting for these errors.Note: unless stated otherwise below, any error in a matched pair will mean thatthe label ("matched") is wrong. Any error in a mismatched pair, even with the person having the wrong identity, will generally be correct (the label of "mismatched" will still be correct).Recep_Tayyip_Erdogan_0004 is incorrect (it is an image of Abdullah Gul). This image appears only in one matched pair in the training set of View 1. Janica_Kostelic_0001 is incorrect (it is an image of Anja Paerson).This image appears in one matched pair in the test set of View 1, and in one matched pair and one mismatched pair (with Don_Carcieri_0001) in fold 1 of View 2.Bart_Hendricks_0001 is incorrect (it is a duplicate image of Ricky_Ray_0001). This image appears in two mismatched pairs in the training set of View 1, and one mismatched pair in fold 2 of View 2. (None of the mismatched pairs are with Ricky_Ray.)Carlos_Beltran_0001 is incorrect (it is a duplicate image of Raul_Ibanez_0001). This image appears in one mismatched pair in the test set of View 1, and one mismatched pair in fold 5 of View 2. (None of the mismatched pairs are with Raul_Ibanez.)Emmy_Rossum_0001 is incorrect (it is a duplicate image of Eva_Amurri_0001). This image appears in one mismatched pair in the test set of View 1 (the mismatched pair is not with Eva_Amurri).Michael_Schumacher_0008 is incorrect (it is an image of Rubens Barrichello). This image does not appear in a matched or mismatched pair, in either view. Reference:Please cite as:Gary B. Huang, Manu Ramesh, Tamara Berg, and Erik Learned-Miller. Labeled Faces in the Wild: A Database for Studying Face Recognition in Unconstrained Environments.University of Massachusetts, Amherst, Technical Report 07-49, October, 2007. 数据预览:点此下载完整数据集。
face标准代码一、概述face标准是一种用于人脸识别的开放源代码库,它提供了人脸检测、人脸识别、人脸比对等功能,广泛应用于安防、金融、医疗等领域。
face标准是由阿里巴巴集团开发的一款开源项目,旨在提供一种简单、高效、可靠的人脸识别解决方案。
二、功能特点face标准代码具有以下特点:1.高效性:face标准采用了先进的算法和优化技术,能够快速准确地检测和识别人脸,大大提高了识别效率。
2.可靠性:face标准经过了大量的实际应用测试,具有很高的稳定性和可靠性,能够应对各种复杂场景下的识别任务。
3.可扩展性:face标准提供了丰富的接口和插件机制,方便用户根据实际需求进行扩展和定制,满足不同场景下的需求。
4.开放性:face标准是一个开源项目,用户可以自由地下载、使用和修改代码,有利于推动人脸识别技术的普及和发展。
三、使用方法使用face标准代码需要以下步骤:1.下载face标准代码库,并解压缩。
2.引入face标准代码库中的头文件和库文件,并进行编译和链接。
3.编写识别程序,调用face标准代码提供的接口进行人脸检测、识别和比对等操作。
4.进行测试和调试,确保识别的准确性和稳定性。
四、示例代码以下是一个简单的示例代码,用于演示如何使用face标准代码进行人脸识别:```c++#include<iostream>#include"facerec/facerec.h"usingnamespacestd;usingnamespaceface;intmain(){//创建人脸识别器对象FaceRecPtrrec=FaceRec::create();//加载人脸图片Matimg=imread("test.jpg");//进行人脸检测和识别vector<Rect>faces=rec->detectMulti(img);vector<string>ids=rec->recognizeMulti(img,faces);//输出识别结果for(inti=0;i<faces.size();i++){cout<<"Face"<<i+1<<":"<<endl;cout<<"Id:"<<ids[i]<<endl;cout<<"Confidence:"<<rec->getConfidence(faces[i])<<endl;}return0;}```五、注意事项在使用face标准代码时,需要注意以下几点:1.确保系统环境符合要求,包括操作系统、编译器等。
gallery set参考图像集Probe set=test set测试图像集face renderingFacial Landmark Detection人脸特征点检测3D Morphable Model 3D形变模型AAM (Active Appearance Model)主动外观模型Aging modeling老化建模Aging simulation老化模拟Analysis by synthesis 综合分析Aperture stop孔径光标栏Appearance Feature表观特征Baseline基准系统Benchmarking 确定基准Bidirectional relighting 双向重光照Camera calibration摄像机标定(校正)Cascade of classifiers 级联分类器face detection 人脸检测Facial expression面部表情Depth of field 景深Edgelet 小边特征Eigen light-fields本征光场Eigenface特征脸Exposure time曝光时间Expression editing表情编辑Expression mapping表情映射Partial Expression Ratio Image局部表情比率图(,PERI) extrapersonal variations类间变化Eye localization,眼睛定位face image acquisition 人脸图像获取Face aging人脸老化Face alignment人脸对齐Face categorization人脸分类Frontal faces 正面人脸Face Identification人脸识别Face recognition vendor test人脸识别供应商测试Face tracking人脸跟踪Facial action coding system面部动作编码系统Facial aging面部老化Facial animation parameters脸部动画参数Facial expression analysis人脸表情分析Facial landmark面部特征点Facial Definition Parameters人脸定义参数Field of view视场Focal length焦距Geometric warping几何扭曲Street view街景Head pose estimation头部姿态估计Harmonic reflectances谐波反射Horizontal scaling水平伸缩Identification rate识别率Illumination cone光照锥Inverse rendering逆向绘制技术Iterative closest point迭代最近点Lambertian model朗伯模型Light-field光场Local binary patterns局部二值模式Mechanical vibration机械振动Multi-view videos多视点视频Band selection波段选择Capture systems获取系统Frontal lighting正面光照Open-set identification开集识别Operating point操作点Person detection行人检测Person tracking行人跟踪Photometric stereo光度立体技术Pixellation像素化Pose correction姿态校正Privacy concern隐私关注Privacy policies隐私策略Profile extraction轮廓提取Rigid transformation刚体变换Sequential importance sampling序贯重要性抽样Skin reflectance model,皮肤反射模型Specular reflectance镜面反射Stereo baseline 立体基线Super-resolution超分辨率Facial side-view面部侧视图Texture mapping纹理映射Texture pattern纹理模式Rama Chellappa读博计划:1.完成先前关于指纹细节点统计建模的相关工作。
中国人颌面研究报告数据库
目前并没有一个特定的数据库专门收录中国人颌面研究报告,但有一些专业学术数据库可以检索相关颌面研究报告,包括以下几个:
1. 中国知网:中国知网是一个综合性的学术资源平台,涵盖了各个学科的期刊、学位论文、会议论文等。
可以通过关键词搜索相关颌面研究报告。
2. PubMed:PubMed是生物医学领域的一个重要数据库,其中包含大量相关颌面研究的文献。
3. ScienceDirect:ScienceDirect是Elsevier出版社旗下的学术资源平台,收录了各个学科的期刊和书籍。
可以通过颌面、口腔等关键词搜索相关的研究报告。
4. MEDLINE:MEDLINE是美国国立医学图书馆(NLM)的一个数据库,包含生物医学领域的大量文献,也包括一些颌面研究报告。
此外,你还可以尝试在各大学图书馆的学术搜索引擎中搜索相关颌面研究报告,或者参考一些学术期刊的目录和专业学会的会议论文集。
USTC-NVIE Database[(natural visible and infrared facial expression database)](由中国科学技术大学安徽省计算与通信软件重点实验室建成并发布,是目前世界较为全面的人脸表情数据库,其中包含大约100名被试三种光照条件下六种表情的可见图像以及长波红外图像,另外表情又分为自发表情与人为表情,人为表情又分为戴眼镜与不戴眼镜两种情况。
为进行(自发+人为)表情识别与情绪分析推理实验提供了充足的实验样本与数据)数据库主页:/发布地址:http://sspnet.eu/2010/08/ustc-nvie-natural-visible-and-infrared-facial-expression-database/■Annotated Database (Hand, Meat, LV Cardiac, IMM face) (Link)■AR Face Database (Link)■BioID Face Database (Link)■Caltech Computational Vision Group Archive (Cars, Motorcycles, Airplanes, Faces, Leaves, Background) (Link)■Carnegie Mellon Image Database (motion, stereo, face, car, ...) (Link)■CAS-PEAL Face Database (Link)■CMU Cohn-Kanade AU-Coded Facial Expression Database (Link)■CMU Face Detection Databases (Link)■CMU Face Expression Database (Link)■CMU Face Pose, Illumination, and Expression (PIE) Database (Link)■CMU VAS C Image Database (motion, road sequences, stereo, CIL’s stereo data with ground truth, JISCT, face, face expressions, car) (Link)■Content-based Image Retrieval Database (Link)■Face Video Database of the Max Planck Institute for Biological Cybernetics (Link)■FERET Database (Link)■FERET Color Database (Link)■Georgia Tech Face Database (Link)■German Fingerspelling Database (Link)■Indian Face Database ([url=/~vidit/IndianFaceDatabase/]Link[/url])■MIT-CBCL Car Database (Link)■MIT-CBCL Face Recognition Database (Link)■MIT-CBCL Face Databases (Link)■MIT-CBCL Pedestrian Database (Link)■MIT-CBCL Street Scenes Database (Link)■NIST/Equinox Visible and Infrared Face Image Database (Link)■NIST Fingerprint Data at Columbia (Link)■ORL Database of Faces (Link)■Rutgers Skin Texture Database (Link)■The Japanese Female Facial Expression (JAFFE) Database (Link)■The Ohio State University SAMPL Image Database (3D, still, motion) (Link)■The University of Oulu Physics-Based Face Database (Link)■UMIST Face Database (Link)■USF Range Image Data (with ground truth) (Link)■Usenix Face Database (hundreds of images, several formats) (Link)■UCI Machine Learning Repository (Link)■USC-SIPI Image Database (collection of digitized images) (Link)■UCD VALID Database (multimodal for still face, audio, and video) (Link)■UCD Color Face Image (UCFI) Database for Face Detection (Link)■UCL M2VTS Multimodal Face Database (Link)■Vision Image Archive at UMass (sequences, stereo, medical, indoor, outlook, road, underwater, aerial, satellite, space and more) (Link)■Where can I find Lenna and other images? (Link)■Yale Face Database (Link)■Yale Face Database B (Link)。
河北农业大学本科毕业论文(设计)题目:基于Eigenfaces的人脸识别算法实现摘要随着科技的快速发展,视频监控技术在我们生活中有着越来越丰富的应用。
在这些视频监控领域迫切需要一种远距离,非配合状态下的快速身份识别,以求能够快速识别所需要的人员信息,提前智能预警。
人脸识别无疑是最佳的选择。
可以通过人脸检测从视频监控中快速提取人脸,并与人脸数据库对比从而快速识别身份。
这项技术可以广泛应用于国防,社会安全,银行电子商务,行政办公,还有家庭安全防务等多领域。
本文按照完整人脸识别流程来分析基于PCA(Principal Component Analysis)的人脸识别算法实现的性能。
首先使用常用的人脸图像的获取方法获取人脸图像。
本文为了更好的分析基于PCA人脸识别系统的性能选用了ORL人脸数据库。
然后对人脸数据库的图像进行了简单的预处理。
由于ORL人脸图像质量较好,所以本文中只使用灰度处理。
接着使用PCA提取人脸特征,使用奇异值分解定理计算协方差矩阵的特征值和特征向量以及使用最近邻法分类器欧几里得距离来进行人脸判别分类。
关键词:人脸识别PCA算法奇异值分解定理欧几里得距离ABSTRACTWith the rapid development of technology, video surveillance technology has become increasingly diverse applications in our lives. In these video surveillance urgent need for a long-range, with rapid identification of non-state, in order to be able to quickly identify people the information they need, advance intelligence warning. Face recognition is undoubtedly the best choice. Face detection can quickly extract human faces from video surveillance, and contrast with the face database to quickly identify identity. This technology can be widely used in national defense, social security, bank e-commerce, administrative offices, as well as home security and defense and other areas.In accordance with the full recognition process to analyze the performance of PCA-based face recognition algorithm. The first to use the method of access to commonly used face images for face images. In order to better analysis is based on the performance of the PCA face recognition system selected ORL face database. Then the image face database for a simple pretreatment. Because ORL face image quality is better, so this article uses only gray scale processing. Then use the PCA for face feature extraction using singular value decomposition theorem to calculate the covariance matrix of the eigenvalues and eigenvectors, and use the Euclidean distance of the nearest neighbor classifier to the classification of human face discrimination.KEYWORDS: face recognition PCA algorithm SVD Euclidean distance目录摘要 (1)ABSTRACT (2)1 人脸识别概述 (4)1.1 人脸识别的研究概况和发展趋势 (4)1.1.1 人脸识别的研究概况 (4)1.1.2 人脸识别的发展趋势 (5)1.2 人脸识别的主要难点 (6)1.3 人脸识别的流程 (6)1.3.1 人脸图像采集 (7)1.3.2 预处理 (7)1.3.3 特征提取 (7)1.4 本章小结 (8)2 人脸图像 (9)2.1 人脸图像获取 (9)2.2 人脸图像数据库 (9)2.3 人脸图像预处理 (10)2.3.1 灰度变化 (10)2.3.2 二值化 (11)2.3.3 图像锐化 (11)2.4 本章小结 (12)3 人脸识别 (13)3.1 PCA算法理论 (13)3.2 PCA人脸识别算法的实现 (14)3.2.1 K-L变换 (14)3.2.2 SVD 定理 (14)3.2.3 PCA算法 (15)3.2.4 人脸识别过程 (16)3.3 程序运行效果 (16)3.4 程序代码 (17)3.4.1 代码类关系 (17)3.4.2 代码的OpenCV相关 (18)3.4.3 关键函数代码 (18)3.5 本章小结 (22)结论 (23)致谢 (24)参考文献 (25)1人脸识别概述1.1 人脸识别的研究概况和发展趋势1.1.1 人脸识别的研究概况人脸识别的研究开始于上世纪七十年代,当时的研究主要是基于人脸外部轮廓的方法。
人脸检索标准
人脸检索标准是指使用人脸图像进行检索时所遵循的一些标准或方法。
以下是一些常见的人脸检索标准:
1. 特征提取:人脸检索的第一步是提取人脸图像中的关键特征,常用的特征提取方法包括局部二值模式(Local Binary Pattern,LBP)、主成分分析(Principal Component Analysis,PCA)
和人脸识别网络(例如基于深度学习的卷积神经网络)。
2. 检索算法:根据提取到的人脸特征,使用相似度度量算法来计算不同人脸之间的相似性。
常见的相似度度量算法包括欧氏距离、余弦相似度和哈希函数等。
3. 数据集:为了进行有效的人脸检索,需要一个包含大量人脸图像的数据库作为检索的基础。
常用的人脸数据库包括LFW (Labeled Faces in the Wild)、Yale Face Database和CASIA-WebFace等。
4. 准确率评估:为了评估人脸检索系统的准确率,通常使用指标如准确率(Precision)、召回率(Recall)和F1-score等来
度量系统在给定数据库上的性能。
5. 应用场景:人脸检索广泛应用于各个领域,包括人脸识别门禁系统、安全监控系统、图像搜索引擎等。
因此,根据具体的应用场景,可以对人脸检索的标准进行定制,以满足特定需求。
需要注意的是,并没有一个统一的人脸检索标准,因为不同的
应用场景和需求可能需要不同的方法和度量指标。
因此,在实际应用中,需要根据实际情况选择合适的标准和方法。
BioID face database(BioID脸部数据库)数据摘要:The BioID Face Database has been recorded and is published to give all researchers working in the area of face detection the possibility to compare the quality of their face detection algorithms with others. During the recording special emphasis has been placed on "real world" conditions. Therefore the testset features a large variety of illumination, background, and face size. Some typical sample images are shown below. The dataset consists of 1521 gray level images with a resolution of384x286 pixel. Each one shows the frontal view of a face of one out of 23 different test persons. For comparison reasons the set also contains manually set eye postions. The images are labeled "BioID_xxxx.pgm" where the characters xxxx are replaced by the index of the current image (with leading zeros). Similar to this, the files "BioID_xxxx.eye" contain the eye positions for the corresponding images.中文关键词:脸部,检测,比较,灰度,大光照变化,英文关键词:face,detection,compare,gray,large illumination,数据格式:IMAGE数据用途:to give all researchers working in the area of face detection the possibility to compare the quality of their face detection algorithms with others.数据详细介绍:BioID face databaseThe BioID Face Database has been recorded and is published to give all researchers working in the area of face detection the possibility to compare the quality of their face detection algorithms with others. During the recording special emphasis has been placed on "real world" conditions. Therefore the testset features a large variety of illumination, background, and face size. Some typical sample images are shown below.Description of the face databaseThe dataset consists of 1521 gray level images with a resolution of 384x286 pixel. Each one shows the frontal view of a face of one out of 23 different test persons. For comparison reasons the set also contains manually set eye postions. The images are labeled "BioID_xxxx.pgm" where the characters xxxx are replaced by the index of the current image (with leading zeros). Similar to this, the files "BioID_xxxx.eye" contain the eye positions for the corresponding images.Image file formatThe images are stored in single files using the portable gray map (pgm) data format. A pgm file contains a data header followed by the image data. In our case the header consists of four lines of text. In detail:the first line describes the format of the image data (ASCII/binary). In our files the text "P5" indicates that the data is written in binary formthe second line contains the image width written in text formthe third line keeps the image height also in text formthe fourth line contains the maximum allowed gray value (255 in our images) The header is followed by a data block containing the image data. The data is stored line per line from top to bottom using one byte per pixel.Eye position file formatThe eye position files are text files containing a single comment line followed by the x and the y coordinate of the left eye and the x and the y coordinate of the right eye separated by spaces. Note that we refer to the left eye as the person's left eye. Therefore, when captured by a camera, the position of the left eye is on the image's right and vice versa.Evaluation of face detection algorithmsTo make it possible to compare the quality of different face detection algorithms on the testset we propose the following distance-based quality measure:Estimate the eye positions with your algorithm and calculate the absolute pixel distance from the manually set positions so that you receive two distance values.Choose the larger value and divide by the absolute pixel distance of the two manually set eye positions so that you become independent from the face's size in the image. We call this value relative eye distance.When calculating this distance for each image you can choose the distribution function of the relative distances to compare some results with others. Alternatively we recommend to rate a face as found if the relative distance is equal to or less than 0.25, which corresponds to an accuracy of about half the width of an eye in the image. The detection rate can directly be calculated by dividing the number of correctly found faces by the total number of faces in the dataset.数据预览:点此下载完整数据集。
颌骨数据采集与处理方法颌骨数据采集与处理方法是口腔医学领域中非常重要的一项工作。
通过采集和处理颌骨数据,可以为口腔医生提供更准确的诊断、治疗和预测数据,同时也可以为口腔疾病研究提供重要依据。
以下将为您介绍颌骨数据采集与处理的具体方法。
颌骨数据的采集主要分为两种方法,即影像学方法和测量学方法。
影像学方法是通过X射线等影像学检查手段来获取颌骨的三维立体影像。
常见的影像学检查包括口腔CT、MRI、数字化X光等。
这些方法对于采集颌骨数据具有很高的准确性和可靠性,可以提供详细的颌骨结构信息。
在进行颌骨数据采集的时候,一定要遵循标准的操作流程,在确保安全的前提下尽可能地减少误差和干扰因素的干扰,以保证数据的准确性和可靠性。
同时,对于不同种类的颌骨,采集和处理的方法也有所不同。
在实际应用中,可以根据具体情况选择最合适的方法进行操作。
颌骨数据的处理通常需要借助计算机技术。
首先需要对采集到的数据进行预处理,包括如何对颌骨进行分割、去除噪声和纠正扭曲畸变等。
然后将处理后的数据导入计算机软件中,进行三维重建和分析。
这些软件通常提供了多种模型处理方法,如模型匹配、特征识别、模型拟合等,可以对颌骨数据进行更精确的分析和定量计算。
在处理颌骨数据的过程中,还需要注意数据的精度、可重复性和可视化效果。
这些因素将直接影响到处理结果和数据的应用价值。
因此,在进行数据处理时,一定要使用高质量的软件、硬件和算法,同时对数据的精度和可靠性进行严格的检验,以保证数据的最大化利用。
总之,颌骨数据采集与处理是口腔医学中不可缺少的一部分。
通过采集和处理颌骨数据,可以提供更准确的诊断、治疗和预测信息,为口腔医学和口腔疾病研究提供更精确的依据。
因此,我们需要选择正确的采集和处理方法,并使用高质量的技术和算法,以确保数据的准确性和可靠性,使其发挥最大的价值。
第36卷第4期2010年4月北京工业大学学报J OURN AL OF BE IJI NG UNI VERS I TY OF TEC HNOLOG YVo.l 36No .4Apr .2010三维人脸衰老数据库的生成司慧琳1,2,孔德慧1,宋彩芳1,尹宝才1(11北京工业大学计算机学院多媒体与智能软件技术北京市重点实验室,北京 100124;21北京工商大学计算机与信息工程学院,北京 100048)摘 要:采用改进的形变模型方法将真实人脸衰老图像重建成三维人脸,进而生成三维人脸衰老数据库,并根据性别和年龄形成三维衰老人脸性别子库和衰老子库.结果表明,此方法可有效地为三维人脸衰老化研究提供三维衰老人脸数据.关键词:衰老人脸;人脸合成;形变模型;三维人脸数据库中图分类号:TP 39119文献标志码:A 文章编号:0254-0037(2010)04-0558-04收稿日期:2009204224.基金项目:国家自然科学基金资助项目(60533030,60825203);国家科技支撑计划项目(2007BA H 13B01);北京市教育委员会科技发展计划面上项目(K M 200710005017).作者简介:司慧琳(1972)),女,江苏阜宁人,博士研究生.三维衰老人脸数据库对衰老预测、年龄估计和衰老人脸重建等三维人脸衰老化研究与应用来说是不可缺少的数据资源.现有的三维人脸数据库按照数据采集的约束条件可构成表情库、姿态库、光照库等,但是没能构成年龄库,这是因为这些三维人脸库中尽管存在少量衰老人脸数据,也大多缺乏准确的年龄信息.另外,与其他三维人脸数据获取相比,大规模获取三维衰老人脸数据存在特殊难度,因为人脸的形状和纹理的衰老变化是在漫长的几十年内(例如从20~70岁)发生的,需要在长时间内对被测对象进行跟踪采集.由此可见,获得同一个人在漫长时期内的三维人脸数据是很困难的.缺乏合适并且足够的三维衰老人脸数据限制了三维人脸衰老化研究.一些三维人脸衰老化研究工作基于几何或肌肉方法合成人脸皱纹,原因是这些研究方法不需要大样本的数据支持,例如:Bando 等[1]在皮肤纹理基础上采用几何方式合成人脸皱纹;Lar bou lette 等[2]依据某形状函数,通过计算皱纹幅值来调整人脸网格顶点的位置合成出人脸皱纹;Wu 等[3]通过肌肉向量来控制人脸的变形合成了带皱纹的人脸.然而采用这些方法合成的人脸衰老细节往往真实感欠佳,在实际应用时备受限制.同期,一些基于统计学习的方法也被运用到三维衰老效果模拟,例如:Choi [4]利用主成分分析(pri n c i p al co mponent ana l y sis ,PC A )从人脸图像提取衰老变化成分(age change co mponents ,ACC)来驱动三维人脸的形状发生变化,从而实现人脸不同年龄阶段衰老效果的模拟;Gol o vi n sk i y 等[5]则对人脸纹理分块进行滤波分析,提取细节特征建立统计模型,然后由统计模型驱动网格变化来合成带衰老细节的人脸,但是基于统计学习的研究方法是否有效在很大程度上受所采用的人脸库的规模和覆盖的人脸数据的影响.为了给三维人脸衰老化的各种研究与比较提供数据平台,作者通过人脸衰老图像提取衰老特征变化合成了三维人脸衰老数据库.1 数据获取构建三维衰老人脸库首先需获取一定数量的衰老人脸数据.直接获取三维衰老人脸数据的途径是采用三维扫描仪,这意味着需要在一段很长跨度的时间内对被测对象进行跟踪采集,难度较大,时间也较长.另外,也可通过间接途径获取三维数据,即通过二维人脸图像重建三维人脸,进而创建具有三维衰老特征的人脸库,这在一定程度上可有效地提供三维衰老人脸数据.第4期司慧琳,等:三维人脸衰老数据库的生成111 基于三维数据扫描仪的数据获取近年来,随着三维数字化扫描仪的应用,直接利用三维数据进行建库日趋增多.国际典型的人脸库是德国Vetter 等[6]使用形变和光流的方法建立的MPI 三维人脸数据库,大约200人规模.国内BJ UT 23D 人脸库是目前最大的中国人三维人脸数据库[7],包括1200名中国人的三维人脸数据,其中500人的数据对外公开发布(男女各250人,年龄分布在16~49岁),人脸数据均是中性表情.112 二维衰老图像的数据获取图像数据的获取比三维数据扫描简单.依据二维图像构建人脸衰老图像库目前已有报道,最为经典的是塞浦路斯大学的FG 2NET 衰老人脸图像库[8],大约有1000张左右人脸图像(主要来自旧照片的扫描)以及详细的年龄信息(年龄段为0~69岁),包括82个人,每个人大约12幅图像,部分图像还提供标定形状的关键点位置数据.FG 2NET 数据库中每个被测对象都具有不同年龄段的多幅人脸图像,有助于人脸数据的建模和仿真.2 三维人脸衰老数据库生成211 基于图像的特定三维人脸重建本文选择FG 2NET 衰老人脸图像库的人脸图像,采用W ang 等[9]提出的三维人脸主动形变模型(3D acti v e morphab le mode,l 3D 2A M M)和一种遗传算法的模型匹配方法不断调整模型参数来实现三维人脸的自动重建,间接得到了三维人脸的衰老数据.该方法与基于多视点照片的经典人脸合成方法也完全不同,它通过数据库中人脸向量的运算来产生新的人脸.该方法是一种改进的形变模型方法,在模型的匹配速度、重建精度和效率方面都得到提高.在三维主动形变模型中,原始三维人脸的形状和纹理分别表示成形状向量和纹理向量,即S i =(X i 1,Y i 1,Z i 1,X i 2,,,X in ,Y in ,Z in )T ,T i =(R i 1,G i 1,B i 1,R i 2,,,R in ,G in ,B in )T (1)其中,1[i [N,N 为三维人脸的个数;S i 为第i 个三维人脸顶点坐标组成的几何形状向量;T i 为由三维人脸顶点对应纹理值(R,G ,B)形成的纹理向量;n 为三维人脸的顶点个数.对三维人脸的形状向量和纹理向量进行线性组合即可产生新的人脸向量S new =E N i a i S i ,T n e w =E N i b i T i (2)式中a i 、b i 为原型人脸的组合系数,且E N i a i =E N ib i =1.基于三维人脸主动形变模型重建出的三维人脸结果如图1所示.图1 基于图像的三维人脸重建结果F i g .1 Or i gi na l i m ages and three 2di m ensi onal reco nstructio n resu lts图1表明,衰老人脸图像中衰老细节(如皱纹、眼袋等)在三维重建人脸中基本得到了保持.这说明由三维主动形变模型重建出三维人脸可作为一种新的真实感三维人脸合成方法来有效地保持细节.559北 京 工 业 大 学 学 报2010年图2 特征点标定F i g .2 Labeli ng of feature poi nts 需说明的是,从FG 2NET 库中选择的图像来自旧照片扫描,其人脸大小、姿态、表情、光照等相差很大,在重建三维人脸前,为了保证所提取的人脸区域基本正确,先从人脸图像中选择代表个人相貌的特征点.因为人脸主要由眼睛、鼻子、嘴巴、下巴等部件构成,这些部件的形状特征具有稳定性,能充分标示人脸区域,所以本文在人脸的上述部件上共选取了9个特征点并通过鼠标在图像上进行了手工标定,如图2所示.图中白色点是手工标定的特征点.特征点的标定使得大小不同、姿态各异的人脸图像在重建三维人脸时取得了一定的矫正,如图3所示.这是因为无论脸部图像大小和姿态差异如何,在重建三维人脸前已经依据特征点的标定做了相应的调整,并且最终生成的三维人脸都具有相同的拓扑结构,有125601个顶点,组合成249856个三角面片.图3 大小和姿态各异的三维人脸重建结果F i g .3 R esu lts of three 2d i m ensi onal face reconstructi on processing d ifferent size and pose212 三维人脸衰老数据库的生成人脸衰老变化通常在20岁之后发生,因此,本文从FG 2NET 衰老人像库中选择人脸图像的依据是该库中某个人脸图像序列中是否具有大于或等于20岁的数据.利用主动形变模型技术将选择的人脸图像序列重建为相应的三维衰老人脸序列,进而生成三维衰老人脸库,具体的流程如图4所示.图4 三维人脸衰老数据库生成过程F ig .4 Constructi on process of t hree 2di m ensio na l aging face database在生成三维人脸库之后,按性别和年龄将三维人脸数据组合形成2个子库:性别子库和衰老子库.其中性别子库包含男性28人,共计342个三维人脸数据,女性22人,共计275个三维人脸数据;衰老子库包含617个三维人脸数据,分别与年龄相对应.3 结束语作者通过人脸衰老图像提取衰老特征变化合成了年龄跨度在0~69岁的三维衰老人脸,并创建了三维人脸衰老数据库.本文所生成的三维人脸数据都是由点的纹理信息、三维坐标信息以及点之间的连接关系构成,并且都被规格化为相同数量的点和三角面片,其拓扑结构完全一致,人脸数据的同一相对位置的点都固定代表同一个面部特征,因此人脸数据具有很好的可控性和表达力.致谢 感谢塞浦路斯大学免费提供FG 2NET 衰老人脸库.560第4期司慧琳,等:三维人脸衰老数据库的生成561参考文献:[1]BANDO Y,KURATATE T,NIS H I TA T.A si m p le me t hod f or modeli ng wr i nk l es on human ski n[C]M P roceed i ng of10t hP ac ific Conference on Co m puter Graphics and App licati ons.Be iji ng:IEEE Co mput Soc,2002:1662175.[2]LARBO UL ET TE C,CAN I M P.R eal2ti m e dyna m i c wri nkles[C]M P roceed i ng of Co mpu ter G raph i cs Interna tiona.l C re te,Greece:IEEE Co mput Soc,2004:5222525.[3]WU Y,T HAL MA NN N M,T HAL MA NN D.A dyna m i c wrinkle m odel i n fac i a l an i m a ti on and sk i n agi ng[J].Jo urna l ofV i sua lizatio n and Co mputer An i m ati on,1995,6(4):1952205.[4]CHO I C.Age change f or pred i cti ng future f aces[C]M1999IEEE Inte rnati onal Fuzzy Syste m s Conference P roceedi ngs.Seo u:lIEEE,1999:160321608.[5]GOL OV I NSKIY A,MAT US I K W,PF IS TER H,et a.l A statistica lm o de l for synthesis of deta iled fac ial geo m etry[J].AC MTransac ti ons o n Graph i cs,2006,25(3):102521034.[6]BL ANZ V,VETTER T.A m orphab le model for the synthes i s of3D faces[C]M Co mputer Graph i cs P roceed i ngs.LosAnge l es:AC M,1999:1872194.[7]YI N Bao2ca,i SUN Yan2feng,WANG Cheng2zhang,et a.l B J UT23D l a rge scale3D face database and i nfor m atio n processi ng[J].Journal of Co m pute r R esea rch and Deve l op m en t,2009,46(6):100921018.[8]FG2NET Consorti um.The FG2NET agi ng database[DB/OL].[2007211212].h ttp:M sti ng.cycoll ege.ac.cy/~a lan iti s/f gnetagi ng/i ndex.ht m.[9]WANG Cheng2zhang,YI N Bao2ca,i SUN Yan2feng,et a.l An i m proved3D f ace modeli ngm etho d based on m orphable model[J].A cta Au t o m atica S i n i ca,2007,33(3):2322239.Constructi on of a Three2d i m ensi onal Agi ng Face DatabaseSI H ui2li n1,2,KONG De2hu i1,SO NG Cai2f ang1,Y I N Bao2ca i1(1.Beiji ng Key Laboratory ofMu lti m ed i a and Intelligent Soft w are Technol ogy,College of Co m puter Science,Be ijing University of Technol ogy,Be ijing100124,Ch i na;2.College of Co mputer Sc i ence and Infor m ati o n Engi neer i ng,Be iji ng Technol og y and Busi nessUniversity,Beiji ng100048,Chi na) Abstr act:A t h ree2di m ensional aging f ace database is i n dispensable data resource f or three2d i m ensi o na l face agi n g research and app lication.In th is paper,usi n g i m proved mor ph mode,l actua l f acial i m ages were reconstructed i n to the correspond i n g three2d i m ensi o na l agi n g fac ial data and f urther into a three2d i m ensional agi n g f ace database,wh ich consists of gender sub2database and agi n g sub2database.The resu lts show that the method is feasi b l e i n supp l y ing three2d i m ensi o na l aging face data.K ey w ord s:age f ace;f ace synthesis;morph mode;l three2di m ensional face database(责任编辑梁洁)。
CBCL FACE DATABASE #1(CBCL人脸数据库#1)
英文关键词:
faces and non-faces,the Center for Biological and Computational Learni,MIT,Recognition,
中文关键词:
脸和非脸部中心生物和计算Learni,麻省理工学院,识别,
数据格式:
IMAGE
数据介绍:
This is a database of faces and non-faces, that has been used extensively at the Center for Biological and Computational Learning at MIT. It is freely available for research use. If you use this database in published work, you must reference:
CBCL Face Database #1
MIT Center For Biological and Computation Learning
/projects/cbcl
CONTENTS:
Training set: 2,429 faces, 4,548 non-faces
Test set: 472 faces, 23,573 non-faces
The training set face were generated for [3]. The training set non-faces were generated for [2]. The test set is a subset of the CMU Test Set 1 [4], information about how the subset was chosen can be found in [2].
The tarballs train.tar.gz and test.tar.gz contain subdirectories with the images stored in individual pgm files. These pgm images can be processed using the functions in the c-code directory. Note that the c-code directory is not a complete program, merely a collection of functions that are useful for reading in the pgm files. These functions have successfully been used numerous times to read in and convert this data; we are unlikely to have time to respond to requests regarding help with these functions, or offer other programming help.
For those who don't want to work with the image files directly, we provide the files svm.train.normgrey and svm.test.normgrey. These files are in the proper format to be read by the SvmFu SVM solver
(/projects/cbcl/), but should be easily convertible
for use by other algorithms. In each file, the first line contains the number of examples (6,977 training, 24,045 test), the second line contains the number of dimensions (19x19 = 361). Each additional line consists of a single image, histogram equalized and normalized so that all pixel values are between 0 and 1, followed by a 1 for a face or a -1 for a non-face. This is the data that was used in [1].
A slightly modified version of the dataset was used in [2] --- the histogram normalization was identical, but additionally, the corners of the image were masked out.
[以下内容来由机器自动翻译]
这是已被广泛用于中心生物及计算学习在麻省理工学院的脸和非脸部的数据库。
它是自由供研究使用。
如果您使用此数据库中发表的作品,你必须引用:CBCL 脸数据库# 1麻省理工学院生物研究中心和计算学习/projects/cbcl内容:训练集:2,429 脸、4,548 非脸测试集:472 脸、23,573 非脸[3] 生成的训练集的脸。
训练设置非-面临的产生[2]。
测试集是一个子集的cmu 中央结算系统测试设置1 [4]、[2] 中可以找到有关如何选择子集的信息。
压缩档train.tar.gz 和只有个别pgm 文件中存储的图像中包含子目录。
可以使用的c 代码目录中的功能来处理这些pgm 图像。
请注
意c 代码目录不是一个完整的程序,只是对于阅读pgm 文件中非常有用的功能的集合。
这些函数成功曾经无数次在读取并转换此数据;我们不可能有时间去响应请求关于帮助这些函数,或提供其他编程帮助。
对于那些不想直接使用的图像文件,我们提供的文件svm.train.normgrey 和svm.test.normgrey。
这些文件被读取
(/projects/cbcl/),SvmFu SVM 规划求解以正确的方式是,但应该轻松自由兑换其他算法使用。
在每个文件中,第一行包含示例24,045 测试6,977 培训),第二行包含的维度(19 x 19 = 361) 数的数。
其他每行包含单个图像,直方图均衡和归一化,使所有像素值都为0 和1 之间的后面的一张脸1 或非工作面-1。
这是[1] 中使用的数据。
[2] 中使用数据集的一个稍加修改的版本---直方图正常化是相同的但此外,图像的边角被遮盖掉。
[自动翻译由Microsoft Translator完成]
点此下载完整数据集。