Acoustic Target Recognition System Research And Implementation

Target recognition is a very important research project in the battlefield. We can find some effective method to beat the enemy when we can identify target correctly. For many years, scientists have suffered a lot of time and effort to research the target recognition. The mostly used way is to find the target by their shape, but it can be easily influenced by their figure disguise, and because we send detection signals, it's likely to uncover ourselves. Recently, scientists are trying to identify the acoustic signal generate from targets to recognize them. The signal is coming from the target itself, so we won't be betrayed because we don't send any signals.Acoustic target recognition is part of the pattern recognition, it's a subject of the automatic target recognition. It includes signal process, feature extraction and artificial intelligence. Artificial neural network is important method in pattern recognition, wavelet transform is very important in signal process. These algorithms are not only applied in target recognition, but also used in other area.Some very important algorithms of modern signal process that be used in acoustic target recognition have been researched in this paper. Both of the theoretical bases and the algorithm implementation have been introduced, and verified the effect in acoustic recognition by MATLAB emulation. We used wavelet transform to decrease noise, implement feature extraction of acoustic signals by AR model and wavelet packets, realized the target classification by BP neural network. The acoustic target recognition system is implemented based on TMS320F2812. The system includes power management, signal filtering, data acquisition and storage, communication module. System control software and the algorithm realization is finished after the hardware is succeed. Experiment results prove, in de-noising environment, the system can finish the signal acquisition and processing, apply identification algorithm, it can give a satisfactory identification result.