Cellular Automata And Self-Organizing Neural Networks For Invariant Target Recognition
Ahmed S. Tolba, Abdul Nasser S. Abu-Rezq · International Journal of Modelling and Simulation · 1999
This paper presents a global and adaptive approach for detection and recognition of moving targets from consecutive binary image sequences. A probabilistic cellular automaton is used for detection of moving targets and a self-organising Kohonen network is used for classification based on the moment in varianta. The proposed approach has many advantages such as suitability for parallel implementation, noise immunity, easy detection of partially occluded targets, and detection of motion in spite of large displacements. Target speed can also be measured by encoding the target displacement information in the cellular automaton. Aircraft detection and classification is used as an application domain to illustrate the performance of the proposed approach.