A Biologically Inspired Hierarchical Neural Network System for LADAR Object Classification

A modular hierarchy of neural networks is proposed as a LADAR ATR (Automatic Target Recognition) system. neural networks perform both the basic segmentation as well as the object classification. HNC and its teaming partner, professor Stephen Grossberg (Director of Boston University's center for Adaptive Systems) propose to develop a LADAR ATR system based upon the earlier work by professor Grossberg et al. in their 1988 publication "Invariant Recognition of Cluttered Scenes by a Self-Organizing ART Architecture: CORT-X Boundary Segmentation". During Phase I of the project a preliminary, non-real time demonstration of the LADAR ATR system will be camed out using simulated LADAR data. This LADAR ATR system will employ the CORT-X and ART (e.g., AKT3 or ARTMAP) neural networks. A design of the Phase II near-real time implementation will also be produced during Phase I.