Abstract: Scattered-wave imaging leverages seismic scattering responses in subsurface heterogeneous media, offering high spatial resolution that is particularly suitable for finely characterizing geological structures such as fractures, karsts, and channels. It demonstrates significant potential in complex carbonate reservoir description and hydrocarbon exploration. However, scattered waves and reflected waves are often mixed in actual seismic wavefields, making effective separation difficult and thereby compromising the imaging accuracy of geological targets. To address this issue, this paper develops a scattered-wave imaging method based on the wave equation. By introducing a spatial structure tensor algorithm to extract stratigraphic dip information from conventional imaging data and using it as a constraint, the method distinguishes between scattered and reflected waves during wavefield separation according to their morphological differences in the angle domain. This technique effectively enhances the contribution of scattered signals in imaging while significantly suppressing reflection interference from complex structural backgrounds. Tests on synthetic data and applications to real data from a complex karst area in southwestern China show that the method maintains good adaptability even in structurally complex regions. It can more clearly highlight the spatial morphology of subsurface heterogeneous anomalies such as caves and collapses, significantly improving the accuracy of scattered-wave imaging and the reliability of geological interpretation. This provides new technical support for high-resolution imaging under complex geological conditions.