A modified reconfigurable super-resolution algorithm for infrared images is put forward on the basis of study and analysis of the Projection Onto Convex Sets (POCS) algorithm. The modified algorithm makes full use of the determination model in the spatial domain and implements registration by smoothing noise reduction processing after motion estimation. Because the Point Spread Function (PSF) is modified while the details of digital images are highlighted, reconfigurable high resolution images with optimal quality are obtained by inhibiting the margin Gibbs phenomenon. The quality of the reconfigured infrared image is evaluated quantitatively. The result shows that the expected result is obtained in image quality. This technology is of important application value to infrared target recognition and tracking, infrared reconnaissance and anti-reconnaissance, study of characteristics of infrared ships, processing of high-definition digital images, ripping of old video and extraction and recognition of biological information.