This paper addresses the limitations of traditional seismic imaging techniques in exploring deep and complex hydrocarbon reservoirs by proposing a least-squares Gaussian beam migration method based on the extraction of angle-domain common image gathers. The method utilizes the angle information carried by the Gaussian beam propagator to separate imaging values at different reflection angles. Combined with the iterative optimization of the least-squares method, high-resolution and amplitude-preserving imaging results are achieved. Furthermore, the concept of principal axis continuation is applied to enhance the amplitude of shallow reflection axes and compensate for the missing signal in the middle and deep layers, thereby further improving the overall amplitude preservation of the stacked section. Model testing verifies the effectiveness of this method, demonstrating that it can more clearly delineate complex structures, effectively remove interference, and enhance the amplitude preservation of the imaging section compared to conventional Gaussian beam migration.