Spectrum sensing feature of cognitive radio systems introduces methods to improve the spectrum utilization. The unique signatures or specific features of wireless signals can be utilized to improve the performance of spectrum sensing by identifying the signals. Moreover, spectral efficiency can be improved by achieving signal orthogonality in different domains such as code, polarization and location beside the frequency and time. However, such methods assume the types of the signals of interest to be known. In this paper, template matching method is proposed as a new spectrum sensing technique to identify the wireless signals based on the spectral signatures in the spectrum. In contrary to the original template matching approach, the number of templates to be kept in the database is limited to one abstract signal for each signal type. The templates are constructed from the abstracts and scaled based on the template matching parameters. Two new metrics are introduced for signal identification method proposed. The performance of proposed metrics are also compared with that of energy detection for wireless fading channels. Identification of the signals with the similar signatures are discussed as well.