RUI: Fractional Power Fringe - Adjusted Filter Based Joint Transform Correlation

项目编号：ECS-9705668项目负责人：Mohammad Alam标题：本科院校研究（RUI）：基于分数功率条纹调整的联合变换相关器在本研究中，将研究一种用于光学模式识别的新型实时畸变不变滤波方法。光学模式识别包括使用基于匹配滤波器的相关器或联合变换相关器（JTC）；只有后者适合实时应用。近年来，人们提出了许多用于光学模式识别的非线性光学技术，其中以全光光折变晶体JTC最为突出。目前，光折变JTC结构仅适用于单目标检测，且处理速度较慢。PI的研究将发展一种新型的广义JTC，该JTC在傅里叶平面上采用非线性解耦。PI的初步研究表明，与其他JTC体系结构相比，这种JTC技术可以获得更好的相关输出。这似乎适用于多个目标，它们可以相对旋转，也可以适用于单个目标；对有噪声和无噪声的背景都适用。这种改进的光学模式识别技术有许多应用：用于工业自动化的机器部件识别和标签检测、光学字符识别、超快速指纹识别、目标检测和跟踪、机器人视觉、自动智能车辆控制系统，以及用于广域搜索情况的实时兴趣区域标识符，如识别地球资源图像中的特定特征、模式或特征。PI将通过广泛的建模、仿真和实验，探索各种架构来有效地实现非线性傅立叶域处理。该实验将与代顿大学的光电实验室合作进行。所有的本科生都计划参加印第安纳州立大学韦恩堡分校和代顿大学的研究。

Proposal Number: ECS-9705668 Principal Investigator: Mohammad Alam Title: Research in Undergraduate Institutions (RUI): Fractional Power Fringe-Adjusted Based Joint Transform Correlator In this research, a novel real-time distortion-invariant filter procedure for optical pattern recognition will be investigated. Optical pattern recognition involves the use of either a matched filter based correlator or a joint transform correlator (JTC); only the later is suitable for real-time applications. Recently a number of nonlinear optical techniques have been proposed for optical pattern recognition, and the all-optical photorefractive crystal JTC stands out among these. To date, the photorefractive JTC architecture is suitable only for single target detection and the processing speed is slow. The PI's research will develop a novel generalized JTC which employs nonlinear apodization at the Fourier plane. Preliminary studies by the PI indicate that this JTC technique results in better correlation output compared to alternate JTC architectures. This seems true for multiple targets, which may be rotated with respect to each other and well as single targets; and for noisy as well as noise-free backgrounds. There are many applications for such an improved optical pattern recognition technique: machine parts recognition and label inspection for industrial automation, optical character recognition, ultrafast fingerprint identification, target detection and tracking, robotics vision, automated intelligent vehicle control systems, and real-time region-of-interest identifier in broad area search situations such as the identification of specific features, patterns or characteristics in earth resource images. The PI will explore various architectures to effectively implement the nonlinear Fourier domain processing, through extensive modeling, simulation and experimentation. The experimentation will be carried out in collaboration with the Electro-Optics Laboratory at University of Dayton. Fu ll undergraduate student participation in the research at INPU-Fort Wayne and at University of Dayton is planned.