Comparison between least-squares reverse time migration and full waveform inversion