Abstract: To address the wavefront curvature error (WCE) induced by the plane-wave approximation in the polar format algorithm (PFA) for high-resolution squinted spotlight synthetic aperture radar (SAR) imaging, this paper proposes a phase-structure-mapping-based full-error compensation method. First, a geometric model for UAV-SAR data acquisition in squinted mode is established, and the two-dimensional spatial-frequency-domain phase expressions after range and azimuth resampling are derived to formulate the WCE model. Then, the structural relationship of the two-dimensional phase error across range spatial frequencies is analyzed. By mapping the azimuth phase error at a reference range frequency point to the complete 2D phase error, the complexity of directly constructing a 2D compensation filter is reduced. On this basis, a spatially variant full-WCE compensation filter is constructed, and local defocus compensation is decoupled from global geometric distortion correction to improve the accuracy for targets at different spatial positions. Simulation results on a point-target array demonstrate that the proposed method effectively suppresses residual defocusing at scene edges, outperforming the conventional second-order SVPF (quadratic phase compensation) in both focusing quality and geometric consistency. Furthermore, a field experiment conducted over a tarmac scene using a UAV-borne millimeter-wave radar demonstrates that the proposed method improves the clarity of edge regions and preserves the geometric consistency of the reconstructed scene, confirming its feasibility for practical SAR imaging.