Abstract: Reflection travel-time inversion is characterized by a low wavenumber for both travel-time and reflection waveform inversion.Moreover, because only travel-time is used instead of amplitude, it can effectively reduce the requirement for accurate reflection in the reverse migration process of traditional reflection waveform inversion.Hence, the number of inversion calculations is significantly reduced.An accurate estimation of the time shift is the key to reflection travel-time inversion.Traditional methods based on channel correlation can effectively estimate global time-shift features; however, it is difficult to accurately estimate the local unsteady time-shift features of seismic channels.In addition, a single-channel estimation can easily result in poor transverse continuity.To enhance the vertical resolution and horizontal continuity of the reflection travel-time inversion, this study considered multidimensional continuity and instability, introduced a low-rank approximation and dynamic image estimation, and constructed a low-rank dynamic matching constraint.The experimental results showed that, compared with the time difference profiles obtained using only dynamic image estimation, the low-rank dynamic matching constraint constructed in this study had better stability in the vertical and horizontal directions, lateral continuity of inverse gradients, consistency with the structural characteristics of geological deposits, and continuity of inversion results.Using the results of this study as the initial model, the full waveform inversion and reverse time-migration imaging results proved the effectiveness of the method.