Spectrum measurement technology is widely used in wireless communication, electronic countermeasure, cognitive radio, smart absorbing structure, etc. In this paper we propose a new sign spectrum sensing approach which contains mainly data acquiring stage and spectrum sensing stage. In the former stage, based on the standard compression sensing, we propose a new sign data acquiring approach, in which we preserve only the sign information about the measurements, so that the data storage space is reduced. In the latter stage, we construct the spectrum sensing algorithm based on the consistent reconstruction principle and the gradient steepest descent algorithm. Simulation results demonstrate that this approach reduces the number of measurements and the data storage space, and at the same time this approach perfectly acquires the spectrum information. This approach can well be used in front spectrum measurement stage of wireless commucation, electronic countermeasure, smart absorbing structure and cognitive radio.