Abstract: Seismic-while-tunneling (SWT) detection tends to encounter challenges including unstable imaging, excessive artifacts, limited target-background contrast, and poor positioning accuracy. To address these issues, this study proposed a dynamic imaging technology based on an improved diffraction-scanning migration imaging (DSMI) method: TA-DSMI. By introducing a true-amplitude weighting function into the DSMI imaging kernel, the TA-DSMI constructs a physically interpretable weighting system consisting of geometric spreading compensation, incident/scattering angle consistency, mild frequency band/absorption compensation, and robust noise suppression. The weighting system is then coupled with an end-to-end process that involves pre-whitening, sliding window stacking, fine-scale scanning of the region of interest (ROI), and local normalization. Consequently, the TA-DSMI achieves minute-level refreshing without significantly increasing the computational overhead. The results indicate that compared to the DSMI, the TA-DSMI improved the contrast-to-noise ratio (CNR) by about 3 dB and reduced the peak positioning error of anomalous bodies by about 1.5 m, with a single imaging time of ≤5 min. Overall, the TA-DSMI significantly enhances the physical consistency and interpretation reliability of SWT dynamic imaging, providing an engineering-feasible technology pathway for real-time advanced identification and spatiotemporal tracking of concealed discontinuities.