A method for performing super-resolution comprises steps of generating the high-resolution, low-frequency spatial and temporal bands of the input video sequence by interpolation, synthesizing the high-resolution, high-frequency spatial band by cross-frame spatial high-frequency extrapolation, and fusing these two bands to generate the spatio-temporally super-resolved video sequence. A corresponding system for performing super-resolution comprises a stage where the high-resolution, low-frequency spatial and temporal bands of the input video sequence is generated by interpolation, a stage where the high-resolution, high-frequency spatial band is synthesized by cross-frame spatial high-frequency extrapolation, and a stage where these two bands are fused to generate the spatio-temporally super-resolved video sequence.