Innovative Concepts for Automated Target Recognition Using Artificial Neural Networks

The need for an automated target recognition system that can identify a target from a noisy image, irrespective of the target's orientation or size is identified. A novel pulse-coupled neural network is proposed to meet this need. The first step is to capture the image. Next, a neural network is used to determine whether an object of interest exists in the image. Next, another neural network determines the class to which the object belongs. Finally, a third neural netwrok identifies the image within that class. In addition, a system to integrate the developed software for each stage will also be implemented. A work plan and a schedule for Phase I is presented and the expected results are detailed. A general outline of Phase II work is also presented. ANTICIPATED BENEFITS: Development of underground and underwater resource identification systems based on ANNs and fuzzy logic, machine vision for automated manufacturing environments, image analysis systems for the medical fields (CAT-Scans, MRI, X-Rays), ANN-based image analysis methods for law enforcement agencies (finger print analysis, hand-print analysis, DNA analysis).