20150429-12 李东: Pore-scale Facial Feature Extraction and Its Application

2015-4-27 18:35| 发布者: 贾伟中科院合肥| 查看: 6034| 评论: 0

摘要: 报告嘉宾2：李东（广东工业大学）主持人：贾伟（中科院合肥物质科学研究院）报告题目：Pore-scale Facial Feature Extraction and Its Application报告时间：2015年4月29日晚20:00（北京时间）文章信息：Dong Li and ...

【15-12期VALSE Webinar活动】

报告嘉宾2：李东（广东工业大学）
主持人：贾伟（中科院合肥物质科学研究院）
报告题目：Pore-scale Facial Feature Extraction and Its Application http://valser.org/webinar/slide/slides/20150429/LiDong_slides.pdf
报告时间：2015年4月29日晚20:00（北京时间）
文章信息：
[1] Dong Li and Kin-Man Lam, Design and Learn Distinctive Features from Pore-scale Facial Keypoints, Pattern Recognition, 48(3), pp. 732-745, 2015. [PDF] 1-s2.0-S003132031400394X-main.pdf
[2] Dong Li, Huiling Zhou and Kin-Man Lam, High-Resolution Face Verification Using Pore-scale Facial Features, IEEE Transactions on Image Processing, 24(8), pp. 2317-2327, 2015. [PDF] 07059198.pdf
报告摘要：Establishing correct correspondences between two faces with different viewpoints has played an important role in 3D face reconstruction and other computer-vision applications. Usually, face images are considered to lack sufficient distinctive features to establish a large number of correspondences on uncalibrated images. In this paper, we investigate pore-scale facial features, which are formed from pores, fine wrinkles, and hair. These features have many characteristics that make them suitable for matching facial images under different variations. Using both biological observation and computer-vision consideration, a new framework is devised for pore-scale facial-feature extraction and matching. The matching difficulty under various skin appearances of different subjects and imaging distortion is also analyzed. For further improving the matching performance and tackling distortions such as varying illuminations and unfocused blurring, a pore-to-pore correspondences dataset is established for training a more distinctive and compact descriptor. Experiments are conducted on a face database containing 105 subjects, and the results prove that the pore-scale features are highly distinctive; face images with a minimum resolution of 600$\times$700 (0.4 mega) pixels contain sufficient details to perform a reliable matching in different poses. Generally, our algorithm can establish between 500 and 2\;000 correct correspondences on a pair of uncalibrated face images of the same person. Furthermore, the proposed methods can be applied to face recognition, differentiating identical twins, 3D reconstruction, etc.
报告人简介：李东，男，广东工业大学讲师。分别于2006年和2009年在天津大学计算机系取得学士学位和硕士学位，2014年在香港理工大学电子信息工程系取得博士学位。研究兴趣包括计算机视觉，模式识别和图像处理。已在 PR, TIP, ICIP, ICME 等优秀期刊和会议上发表论文多篇。首次提出利用双眼间距大于280像素的人脸图像提取可重复定位检测的毛孔尺度特征，从而成功进行了无标定人脸图像毛孔到毛孔的一一匹配，进而提出了一种利用毛孔尺度特征的高清人脸识别方法。近期，利用改进的毛孔尺度特征，在ND-Twins数据库可控条件下取得了99.8%的识别率，不可控条件下依然取得超过95%的识别率，为当前世界最好成绩。更多信息请见http://drdongli.github.io。

收藏邀请

上一篇：VALSE2015欢迎您的到来！下一篇：20150429-12 廖胜才: 不完整人脸无配准识别

20150429-12 李东: Pore-scale Facial Feature Extraction and Its Application

最新评论

相关分类