Abstract: Compared with time-domain seismic data, depth-domain counterparts have significant advantages in imaging areas with complex geological structures and strong lateral heterogeneity. Depth-domain horizon interpretation data are being applied to comprehensive reservoir prediction. Time-domain horizon interpretation data are relatively mature in most work areas. To quickly obtain depth-domain horizons, conventional methods mainly convert time-domain interpretations to the depth domain based on the time-depth relationship. Due to inaccuracies in this relationship, the resulting depth-domain interpretations often suffer from positioning errors and horizon misalignment. Therefore, this paper proposes a time-depth dual-domain horizon mapping method based on dynamic image warping. By identifying similar structural features between time-domain and depth-domain seismic data, this method establishes a 3D conversion factor that enables mutual mapping of interpreted horizons across the two domains. This method employs dynamic programming to establish a similarity matrix, which is then used for dynamic accumulation planning to obtain a spatial similarity matrix volume. The conversion factor is subsequently generated by backtracking along the optimal path within this volume, and finally used to map horizons between the time and depth domains. A field-data test in an eastern China oil field shows that the method achieves accurate matching of horizons from seismic data acquired in different periods and different domains, and can accurately cross faults with large throw. Compared with conventional automatic horizon tracking methods, the proposed method shows a great improvement in accuracy.