In order to improve the quality of remote sensing image reconstruction in spatial domain, point spread function of improved projection onto convex set (POCS) algorithm was adopted and an improved POCS super-resolution reconstruction algorithm was proposed. Firstly, the basic principle and implementation steps of POCS algorithm were given. On this basis, the algorithm was improved and the reconstructed high-resolution initial frames were detected on edge. The improved point spread function (PSF) was applied to the detected edge pixels. The horizontal and vertical direction coefficients of PSF corresponding to the pixels at the edge were set with different weights according to the change of the slope of the edge. Finally, two sets of data sets were used to verify the effectiveness of the improved POCS algorithm. The results show that the improved POCS algorithm effectively improves the effect of image reconstruction. The average absolute errors of two groups increase by 0.79% and 0.26%, respectively. It achieves the goal of improving the quality of image reconstruction. The algorithm has good practical application value.