When implementing ultra-high resolution in the device grouping model, Kotelnikov's used theorem on image discretization and kernels of integral image deconvolution operators is aimed at spatial frequencies that lie far enough beyond the limits of transparency of the spatial frequency-contrast characteristic of the area-sensing virtual path. To further compensate for the loss of sharpness in the virtual channel due to dispersions in multiple overlay operations with images, the recursive method of extending the spatial-frequency spectra of images is implemented. The modified super-resolution method Iterative Back Projection for a group of images with a resolution of 1 m on the ground received patterns with a resolution of 0.5 m and further images were obtained with support for a resolution of 0.25 m (estimated by the Foucault method), which corresponds to the level of resolution for devices US intelligence. For a satellite with a resolution of 30 cm, support of 15 cm is obtained.