Technology Research, Target Identification Based On Neural Networks

This paper based on TV guidance studies image automatic target recognition for TV seeker, and offers theoretical preparation and technology support to the development of weapon system.First of all, according to the feature and technology requirements of target recognition for TV guidance, we study neural networks' structure and algorithmic character, and adopt BP Neural Networks to build recognition system for seeker.Secondly, an automatic recognition system frame is designed with confirming network structure. Some system parameters and improved algorithms of BP Neural Network are tested and compared. The efficiency of recognition system greatly is improved through ameliorating parameters of network and program methods. The system memories targets through learning, and simulates parallel processing of the brain. It recognizes targets quickly no matter what the pose, size and rotation of targets are.Finally, target recognition software is developed by VC++, and results are satisfied. It's shown that it's feasible of neural network technology to realize recognition for TV seeker. The system is the base for further research.