Spectrum sensing technology is one of the core technologies of cognitive radio. As the future development of wireless communication technology requires highspeed data communication, wideband spectrum sensing technology has become the focus of the current research. Wideband can’t be directly divided into occupied or idle, so it needs to be subdivided. The signal is divided into multiple subbands, and the problem of multivariate classification is transformed into binary classification problem by pretreatment. In order to reduce the complexity of the spectral sensing algorithm, a spectrum sensing algorithm based on estimation of noise(EN) and support vector machine(SVM) is proposed. The algorithm uses a slow sensing algorithm as estimation of noise and then combines it with low fast sensing algorithm for spectrum sensing. The experimental results show that the proposed algorithm has a significant improvement in the detection performance under the low SNR and can fully recognize the usage of each subchannel under SNR of -10 dB.