As different types of noises coming from different vehicles and weapons mix together, it is a challenge to identify acoustic objectives in such a noisy situation. A systematical study of acoustic objective recognition in wireless sensor network is conducted in this paper. At first, the acoustic target recognition model is introduced, which is based on the characteristics of wireless sensor networks aiming to identify the various target signals in a noisy battlefield. Secondly, the feasibility of acoustic feature extrication by using wavelet packets in the system is analyzed. Then, neural network classifier and the maximum likelihood classifier are compared concerning their advantages and disadvantages respectively. Finally, experimental results are illustrated and they show that: (1) the wavelet packet feature extraction algorithm is feasible and effective, (2) both maximum likelihood algorithm and neural network algorithm can be used for acoustic objective recognition in wireless sensor networks, even neural network classifier usually has higher recognition rate.