Effective extraction of transient impulses from harmonic signals is very important to improve the efficiency of power system analysis. Aiming at the defects of traditional fast Fourier transform (FFT) in detecting the transient impulses buried in the harmonic signals, a novel signal detection method based on union dictionary is proposed. The original separation and detection problem can be converted into a constrained optimization problem by applying the morphological component analysis (MCA). Through combining the union dictionary and alternating direction method of multipliers (ADMM), the constrained optimization problem can be solved. Simulation results show that the proposed method can not only effectively separate the transient impulses and the harmonic signals, but also can detect the features of them precisely.