The quality factor () is an important parameter for measuring the attenuation of seismic waves. Reliable estimation and stable inverse filtering are expected to improve the resolution of seismic data and deep-layer energy. Many methods of estimating are based on an individual wavelet. However, it is difficult to extract the individual wavelet precisely from seismic reflection data. To avoid this problem, we have developed a method of directly estimating from reflection data. The core of the methodology is selecting the peak-frequency points to linear fit their logarithmic spectrum and time-frequency product. Then, we calculated according to the relationship between and the optimized slope. First, to get the peak frequency points at different times, we use the generalized S transform to produce the 2D high-precision time-frequency spectrum. According to the seismic wave attenuation mechanism, the logarithmic spectrum attenuates linearly with the product of frequency and time. Thus, the second step of the method is transforming a 2D spectrum into 1D by variable substitution. In the process of transformation, we only selected the peak frequency points to participate in the fitting process, which can reduce the impact of the interference on the spectrum. Third, we obtain the optimized slope by least-squares fitting. To demonstrate the reliability of our method, we applied it to a constant model and the real data of a work area. For the real data, we calculated the curve of the seismic trace near a well and we get the high-resolution section by using stable inverse filtering. The model and real data indicate that our method is effective and reliable for estimating the value.