Abstract: ‍ ‍Severe layover occurs when ground-based interferometric radar is applied to building imaging as a consequence of insufficient elevation resolutions. Tomographic synthetic aperture radar technology offers elevation resolutions and can realize the three-dimensional （3D） imaging of buildings. Ground-based tomographic arc-scanning synthetic aperture radar （GB-TomoArcSAR） controls the antenna to perform circular scanning in the horizontal plane of different pitch angles through a two-axis turntable to obtain the synthetic aperture in the elevation direction and realize 3D tomography. In this study， a 3D point-cloud generation method for GB-TomoArcSAR is proposed. First， a tomography geometric model suitable for arc sampling in the height direction is constructed. Second， a singular value decomposition method based on the Butterworth filter is used to estimate the spectrum， obtain the peak value in the tomography spectrum and its corresponding peak position， and form the candidate set for the tomography target. Subsequently， a sequential generalized likelihood ratio test with a cancellation detector is performed to estimate the number and position of scatterers， whereas the peak value of the actual target and the corresponding peak position are selected from the candidate set by setting the detection threshold. Finally， a point-cloud optimization method based on spatial geometric distribution is used to eliminate error points and generate a point-cloud image. Simulation experiments involving point and surface targets demonstrate the suitability of the proposed method for GB-TomoArcSAR and its effectiveness in solving the layover of multiscatterer targets in the elevation direction. Additionally， the measured data are verified. Based on the proposed method， the tomographic point cloud of a building foundation pit in Beijing is obtained， which shows consistency with the geometric characteristics of the actual scene.