Abstract: Since no targeted seismic inversion techniques are available for the prediction of complex sandstone-mudstone reservoirs, conventional inversion methods struggle to provide accurate sand body predictions in the presence of sandstone-mudstone interbeds or horizontal fractures. Hence, this study selected Poisson's impedance as the sensitive parameter for reservoir prediction based on the petrophysical parameter cross plots using log data from the study area. Considering the characteristics of fracture development in the reservoirs, this study derived the equation of anisotropic reflection coefficients in vertical transverse isotropy (VTI) media, incorporating Poisson's impedance. Furthermore, this study designed a forward modeling operator using a non-stationary convolution model to improve the strong attenuation characteristics of seismic waves generated by sandstone-mudstone interbeds and protect the reflection characteristics of deep thin reservoirs. Finally, this study proposed a high-resolution direct inversion of Poisson's impedance and fracture parameters using prestack seismic anisotropy data, providing a novel approach for high-resolution characterization of the distribution of sandstone-mudstone reservoirs and the assessment of fracture-bearing properties. Both model and field data tests validate the feasibility and adaptability of the proposed method.