Identification and tracking of dim moving targets in FLIR using artificial neural networks

One of the important challenges in automatic target recognition system is to identify and track slow moving targets under heavy background clutter and noise. The problem becomes more complicated when the moving target is dim and spatially distributed over only a few pixels. In this paper, we propose a novel technique based on artificial neural networks (ANNs) for the identification and tracking of dim moving targets in forward looking infra-red (FLIR) imagery. The ANN is used in a neural adaptive line enhancer (NALE) configuration to identify and detect the dim moving target. First, we used a multilayer perceptron (MLP) for this purpose. The performance of the MLP is found to be satisfactory. Next, we introduced a computationally efficient single layer functional link ANN (FLANN). Performance comparison of the two ANNs was carried out through extensive computer simulation and showed that the FLANN provides satisfactory result with substantial reduction in computational complexity.