Abstract: In the identification of illegal mining issues such as unlicensed extraction and boundary (layer) excavation of underground minerals, efficiently and accurately acquiring spatial information about underground workings is a prerequisite for distinguishing between legal and illegal mining. Traditional methods for regulating illegal mining face challenges such as low efficiency, poor accuracy, and weak reliability. To address these issues, explores methods and their feasibility for inverting the spatial information of underground mining faces using measured surface deformation data, leveraging the relationship between underground extraction and surface deformation characteristics. Targeting the current limitation that existing methods can only be applied to rectangular mining faces, an accurate positioning method for irregular working face is proposed based on real measured surface deformation field data. This method divides the inversion localization problem of underground workings into two parts: the inversion of geological parameters and the inversion of the plane position of the working face. It utilizes a pattern search algorithm to invert the geological parameters, while integrating a morphological algorithm capable of inverting the corresponding plane grid locations of the working face, thereby achieving a unified inversion of geological parameters and the plane boundary of the working face. The reliability of the method is validated by constructing a full-basin point cloud deformation field dataset. The results show that the algorithm achieves high inversion accuracy, with errors in the mining thickness, mining depth, dip angle, and dip direction being 0.03 m, 6.54 m, 0.35°, and 1.04°, respectively; the plane boundary error of the working face is less than 20 m, the corner point plane error is less than 60 m, and the corner point mining depth error is less than 30 m. The algorithm demonstrates stability, able to withstand various factors such as point density, monitoring data noise, and localized voids in point cloud data, thus meeting engineering localization accuracy requirements. A case study on the inversion of the location of the irregular working face at 73_upper 27 further validates the reliability of the proposed method. The research findings also provide reference and insights for surface subsidence analysis and investigations into the mining history of abandoned mines.