Compressed sensing 3D ISAR imaging method integrating multi-angleobservations and a physical scattering model embedding

Traditional inverse synthetic aperture radar imaging suffers from the loss of three-dimensional (3D) target information and geometric distortion caused by fixed rotation axes in single-view observations. To address these limitations, this study proposes a multi-angle, physically precise enhancement method. By constructing 80 collaborative observation perspectives to achieve full-angle coverage, structural occlusion is resolved using 3D point-cloud fusion and registration techniques. By linking the rotation center to the radar line of sight in real time, rather than using a preset fixed axis, the rotation center offset error is fundamentally eliminated while a motion-imaging coupling mechanism suppresses cross-resolution unit migration. In addition, a parameterized target-component scattering model embedded with physically accurate differentiated radar cross-section characteristics is introduced to ensure feature consistency through multi-channel joint reconstruction. Application results show that this method improves target-structure recognition to 92%, increases 3D reconstruction efficiency by 40%, and reduces geometric distortion by 62%, thereby meeting the requirements of high-precision military target recognition.