In this paper, we propose a fast and accurate radar imaging algorithm that combines Kirchhoff migration with Stolt's frequency-wavenumber (F-K) migration. F-K migration is known as a fast imaging method in the F-K domain, while Kirchhoff migration is reported to be more accurate. However, Kirchhoff migration requires the reflection points to be located as a function of the antenna position and the delay time. This prevents the use of fast Fourier transforms because Kirchhoff migration must be processed in the time domain, and this can be extremely timeconsuming. The proposed algorithm overcomes this hurdle by introducing the texture angle and the inverse boundary scattering transform. These two tools enable the locations of the reflection points to be determined rapidly for each pixel of a radar image. The radar signals are then modified according to the Kirchhoff integral, before Stolt F-K migration is applied in the frequency domain to produce an accurate radar image. To demonstrate the performance of the proposed method, the conventional delay-andsum (DAS) migration, Kirchhoff migration, Stolt F-K migration, and the proposed method are applied to the same measured datasets.