De-noising nonstationary signal based on sparse representation and particle swarm optimization

It is difficult and important to de-noise nonstationary signal. To this end, a new noise attenuation method for nonstationary signal was proposed based on sparse representation and Particle Swarm Optimization(PSO). A redundant dictionary which is insensitive to useful signal was developed for the representation of cultural noises. PSO was used to improve the search strategy of Matching Pursuit(MP). Simulated experiments and real MT data were used to test the proposed scheme. As a conclusion, not only charge-discharge-like noise can be effectively removed, spikes and some other irregular noise can also be well suppressed. The apparent resistivity and phase curves obtained after applying our scheme are greatly improved over previous.