Adaptive Fourier single-pixel imaging based on directional energy continuity in high frequencies.

• Proposed a new sampling method for Fourier single pixel imaging, by which significant Fourier coefficients were effectively sampled at low sampling ratios, so the image resolution is greatly improved. • This method innovatively takes the use of the property that the Fourier domain images of most natural sceneries have light-rays pattern and the light rays corresponds to the borders in spatial domain. Therefore the discrete and sparse light rays enable effective sampling of high frequencies that are important for image resolution. • An analysis of randomly chosen 200 images from VOC image database proved most natural scenery images have the light rays pattern in Fourier domain. • The proposed method provides a potential direction to improve the sampling efficiency of Fourier single-pixel imaging. Fourier single-pixel imaging (FSI) technology has attracted wide attention for its high imaging efficiency and broad applicability. However, the existing sampling methods cannot effectively sample the key high-frequency Fourier coefficients with low sampling ratio. To address this problem, we propose an adaptive sampling method based on spectral energy continuity, especially in the directional distributed frequencies that correspond to the borders in the image. In this method, the significant high-frequency components are predicted from simply isometric sampled low frequencies and the generated sampling trajectory is close to the optimal. Because more important high-frequency information is sampled, the image quality at low sampling ratio is greatly improved. We demonstrate with both simulations and experiments that this sampling method is effective in suppressing noise and obtaining detailed information of the image at low sampling ratios. [ABSTRACT FROM AUTHOR]