Abstract: The geological hidden disaster-causing bodies such as small-scale faults and collapsed columns may destroy the continuity of the coal seam, which will lead to accidents such as gas outburst and water outburst in the air mining area, and seriously threaten the coal mining safety.Traditional seismic exploration methods are mainly involved with reflected waves that have a limited resolution and cannot effectively identify them.A multi-parameter sparse optimization diffraction separation method in the shot domain is proposed that is based on curvelet sparse transformation and plane-wave decomposition.It can solve the problem of diffraction separation in the interfering or tangency cases and the separated diffractions have a waveform consistency and integrity.The 3D numerical modeling and simulation test results show that the proposed method can effectively eliminate the shielding effect of strong reflections, separate the complete diffracted/scattered wave, and improve the imaging accuracy of fault edges and scattering points, where the imaging features of polarity reversal can be used to distinguish different types of small-scale geological structures.The field data of Shanxi Yang Erkuang coal mine confirmed that the proposed method can effectively remove the shielding effect of strong reflections from coal seam, and the hyperbolic forms of diffractions in the stacked profile can qualitatively control its reliability.For seismic diffraction interpretation, full-wave field imaging results are generally required to provide a macroscopic geological background and the diffraction imaging results are combined to reveal hidden disaster-induced geological bodies.The diffraction image of collapsed column boundaries and their insider characterization is clearer, and the exhibition of small-scale fractures is more powerful.It is a high-resolution imaging method and has a great potential in detecting hidden disaster geological bodies, which can provide a technical guarantee for coal mine safety and efficient mining in China.