Synthetic aperture radar (SAR) 3-dimensional (3D) imaging uses multichannel systems to acquire 3D ground object information, overcoming the limitations of traditional 2D imaging. Because of its convenience and versatility, SAR 3D imaging is widely applied on unmanned aerial vehicle (UAV) platforms. However, the low altitude of UAV-borne systems introduces geometric inconsistencies in the observation of the same target across different channels. This channel migration challenges the coregistration process in traditional tomographic SAR or array interferometric SAR (array-InSAR) methods, leading to inaccuracies in the reconstruction results. This paper presents a novel framework that fundamentally addresses this issue by integrating 3D back projection with compressive sensing (CS) for typical urban environments, where scatterers are sparsely distributed within each resolution cell. First, a 3D array-InSAR imaging model for low-altitude scenarios is introduced, utilizing a hybrid polar-Cartesian coordinate system. Subsequently, an interpolation method for 3D grids is introduced to replace coregistration, reducing runtime and eliminating multipath signals strategically. Finally, using the sparsity of the scene, the 3D information is reconstructed via CS with an expanded sensing matrix. The effectiveness of the proposed method is validated through simulated and real-data experiments under the sparsity assumption. Results demonstrate a remarkable enhancement in reconstruction quality, with the peak signal-to-noise ratio and normalized root mean square error improving by over 50% and the 3D entropy metric increasing by approximately 0.1.