Abstract: In ultra-wideband radar imaging applications, there exist non-orthogonal sidelobes which would affect the image. The relationship between point spread function and spatial spectrum supporting region is analyzed in this paper. To improve the image quality, an approach based on virtual sub-aperture radar images is proposed to reduce the non-orthogonal sidelobes. This approach separates the mainlobe from the sidelobes, which suppresses the sidelobes and keeps the mainlobe simultaneously. The simulation results show significant reduction of the sidelobe level.