Full Waveform Inversion (FWI) can provide an accurate velocity modelby matching observed and simulated seismograms. Mathematically, FWIis a highly ill-posed inverse problem that the inversion results lackdependence on the observed data. For a stable and reasonable inversion,proper regularization methods have to be taken into account.We proposea novel composite regularization for frequency-domain FWI problem,which uses a non-convex second-order total variation (TV) term and anoverlapping group sparse TV (OGS-TV) regularization term. Comparedwith the conventional TV regularization, our method has better accuracyand robustness, and could effectively make use of the structuralsparsity in the velocity model. Furthermore, the alternating directionmultiplier algorithm with the adaptive selection of the penalty parametersis developed to solve this composite constraint problem, which canimprove the stability of the FWI process. To illustrate the superior performanceof the proposed FWI method both visually and quantitatively,we present several numerical examples through the comparison betweenour FWI method and conventional FWI method with TV regularization.