SBIR Phase II: Robust, Intelligent and Practical Face Recognition Based on Optical Joint Transform Correlation and Neural Networks

SBIR第二期：基于光学联合变换相关和神经网络的鲁棒、智能、实用的人脸识别

• 批准号: 9983308
• 负责人: Andrew Kostrzewski
• 金额: $ 40万
• 依托单位: PHYSICAL OPTICS CORPORATION
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2000
• 资助国家: 美国
• 起止时间: 2000-06-01 至 2002-05-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9983308&HistoricalAwards=false
• 关键词: SBIR; Phase; II; Robust; Intelligent

This Small Business Innovation Research Phase II project will enable Physical Optics Corporation (POC) to build a highly robust and adaptive face recognition system called the Opto-electronic Intelligent Face Recognition System (OIFRS). This system will uniquely combine an artificial neural network and an optical joint transform correlator (JTC) to increase invariance to distortion, shift, scale, and facial expression. In Phase I, we conducted a concept feasibility study and demonstration, developing an optical face recognition system. In Phase II, POC will implement a parallel optical JTC to increase processing speed, accommodate a large number of training patterns, and enhance scalability. POC will design, develop, and implement an innovative, adaptive, nonlinear, self-aligned pseudo-phase-conjugate joint Fourier-Fresnel transform correlator, which will make recognition highly invariant to longitudinal translation and tilt of the input face, and to head position. The proposed innovative OIFRS technology will lead not only to a new robust face recognition system but also to a variety of other security systems with high dual use application potential.OIFRS versatility will benefit not only the national security concerns of DOE, DoD, and other agencies that require highly secure access control, but also commercial access control, and credit card verification.

这个小型企业创新研究第二阶段项目将使物理光学公司（POC）建立一个高度鲁棒和自适应的人脸识别系统，称为光电智能人脸识别系统（OIFRS）。该系统将独特地结合联合收割机的人工神经网络和光学联合变换相关器（JTC），以增加不变性的失真，移位，规模，和面部表情。在第一阶段，我们进行了概念可行性研究和演示，开发了一个光学人脸识别系统。在第二阶段，POC将实现并行光学JTC，以提高处理速度，容纳大量的训练模式，并增强可扩展性。POC将设计、开发和实现一种创新的、自适应的、非线性的、自对准的伪相位共轭联合傅里叶-菲涅耳变换相关器，这将使识别对输入面部的纵向平移和倾斜以及头部位置高度不变。OIFRS的多功能性不仅有利于DOE、DoD和其他需要高度安全访问控制的机构的国家安全问题，而且有利于商业访问控制和信用卡验证。