Abstract: The quality factor Q of subsurface media quantifies seismic energy attenuation and is pivotal for reservoir characterization, hydrocarbon assessment, and subsurface imaging. The spectral-ratio method is one of the most widely used approaches for estimating Q; however, its accuracy degrades markedly in low signal-to-noise ratio (SNR) and for nonstationary signals, limiting its applicability to reservoir prediction. To address these issues, we propose a Q-estimation method that integrates an improved S-transform with Shaping regularization. First, we obtain a high-fidelity time–frequency spectrum by leveraging a nonlinear window-scaling factor introduced in the improved S-transform. Next, we reformulate the division inherent in the spectral-ratio method as an inversion-like problem. Within the Shaping-regularization framework, we construct a two-dimensional triangular smoothing operator that imposes continuity priors jointly along the spatial and frequency directions. Through iterative updates, the procedure suppresses random noise and isolated outliers, yielding accurate and stable Q estimates. Tests on noisy synthetic models demonstrate that the method remains robust under low SNR. Applications to field seismic data show that the estimated Q matches well with water saturation curves and accurately delineates thin tight-sandstone reservoirs. The proposed method provides a reliable and efficient solution for Q extraction and reservoir prediction under low-SNR conditions.