SBIR Phase I: 3D Modeling Facial Aging Effect for Enhancing Facial Recognition Performance

这项小型企业创新研究第一阶段研究项目将开发一种新的三维面部建模方法，以模拟人类受试者的面部衰老效应，以提高面部识别性能。尽管面部识别技术取得了许多进展，但现有系统的性能仍然对注册图像和探测图像之间的某些变化非常敏感。这些因素包括姿势、灯光和表情。最近由16个美国政府机构执行的2002年面部识别供应商测试(FRVT)定量记录了另一个重要因素：它显示，在拍摄受试者图像的时间和系统捕获新图像和执行识别的时间之间，由于面部变化，性能每年下降5%。SBIR提案将开发一种独特的3D人脸建模和计算机化的年龄递增/回归技术来解决这一问题。这项独特的技术完全开发后，能够消除数据库中老化图像造成的性能恶化，并使面部识别系统能够在现实世界的应用程序中进行可靠的现场部署。拟议的产品可能会对国土安全部和其他安全相关应用程序产生重大影响。地方、联邦和国际执法组织将受益于能够识别出他们很长一段时间没有“见过”的人。

This Small Business Innovation Research Phase I research project will develop a novel three-dimensional facial modeling approach to simulate the facial aging effect of human subjects for enhancing facial recognition performance. Despite many advances made in facial recognition technologies, existing system performance is still very sensitive to certain changes between the enrollment image and probe images. These factors include pose, lighting, and expression. The recent Facial Recognition Vendor Tests (FRVT) 2002 performed by 16 US Government agencies quantitatively document another important factor: It shows a 5% drop of performance per year due to the facial changes between the time when a subject's image is taken and the time when the system captures a new image and performs recognition. The SBIR proposal will develop a unique 3D face modeling and computerized age progression/regression technique to solve this problem.This unique technology, when fully developed, is able to eliminate performance deterioration due to aged images in the database and enable robust, field deployments of facial recognition systems in real-world applications.The proposed product could have a big impact with Homeland Security as well other security related applications. Local, federal and international law enforcement groups will benefit from being able to identify people they have not "seen" in a long period of time.