In order to solve the problems of high sampling rate which could shorten the lifetime of wireless sensor networks in wild relics security system, the power spectrum reprocessing is applied to the feature extraction of seismic wave signal. This method is used to solve the target recognition problem, and three data sets are used to verify its effectiveness. The result shows that the power spectrum reprocessing can reduce the communication consumption and prolong the lifetime of networks. This method can also increase the accuracy of target classification. This method, indicating its good performance, had been applied to wild relics security system of Museum of the Terra-cotta Warriors and Horses of Qin Shihuang.