Large aperture static interference imaging spectrometer (LASIS) is characterized by that all CCD chips of 180 spectral bands scan through the same field. Based on this, a novel method is proposed to achieve high accuracy band-to-band registration by re-imaging with virtual CCD based calibration. Considering the geometric factors that cause hyperspectral image mismatch, such as changes of attitude and orbit data, lens distortion and so on, this method reimages the hyperspectral images with the virtual CCD calculated from the calibrated interior orientation to remodel the error sources and eliminate the relative errors. Validation experiments using different area data of RS-14 show that with the proposed method, the band-to-band registration accuracy of LASIS is better than 0.16 pixels.