US-Bangladesh/Kuwait Cooperative Research: Real Time Pattern Recognition

University Of Alabama Tuscaloosa, Tuscaloosa AL

Investigators

Abstract

0002107 Alam Description: This award is for support of a cooperative project by Dr. Mohammad Alam, Department of Electrical and Computer Engineering, University of Alabama, Tuscaloosa, Alabama, Dr. A. K. Cherri, Electrical Engineering Department, Kuwait University, Kuwait and Dr. M.N. Islam, Electrical Engineering Department, Bangladesh University of Engineering and Technology, Dhaka, Bangladesh. These scientists plan to develop a highly efficient joint transform correlator (JTC) based pattern recognition system for real time applications such as finger print identification, adaptive face recognition and classification of weeds from crops. A JTC is inherently suitable for real-time matching and tracking operations since no complex filter is needed. Among the various JTC techniques proposed, the finger-adjusted JTC, which employs a nonlinear finger-adjusted filter at the Fourier plane, has been found to yield the best correlation output for target (pattern) recognition. However, the problem becomes more challenging if the input scene contains target-like clutter, targets that are rotated and have scale variations with respect to the reference, and are affected by other type of 3D distortions. In this project the PIs plan a novel generalized nonlinear JTC, for distortion invariant detection and identification both single as well as multiple targets. Adaptive learning will be incorporated to update the reference image used in the proposed JTC. Scope: The tasks in this project will be carried out by the scientists in the three countries according to their special expertise and the technical resources and instrumentation available to each. Graduate students from the two foreign countries are expected to participate in this international collaboration. Drs. Alam, Cherri and Islam had previously interacted in dealing with research on optics and on electronics and their applications. The planned research will build upon this linkage and is likely to help the three countries in solving certain problems of personnel identification in banking, customs and immigration and in object identification in remote sensing applications in agriculture. This proposal meets the INT objective of encouraging collaboration by US and foreign scientists in areas of potential mutual benefit.