In GHz-level wideband spectrum sensing, too high sampling rate exceeds the specifications of existed analog-to-digital converters (ADC) by sampling directly. Moreover, an accurate estimate of occupied spectrum band can improve spectrum utilization. Therefore, with modulated wideband converter (MWC), a wideband spectrum sensing approach based on multiband signal sampling and wavelet transform is proposed in this paper. With MWC, sub-bands signals are firstly obtained by low rate sampling; and then, an estimation method for noise power and threshold is proposed, such that the spectrum sensing for signal sub-bands can be carried out by energy detection; at last, the edge detection for signal sub-bands by using wavelet transform is used to obtain the exact locations of spectrum occupied by primary users. Simulation results show that the proposed wideband spectrum sensing approach is feasible and effective.