Abstract: BNU-1, the first civil polar optical micro-sat developed by Chinese scientists, is equipped with a medium and high-resolution multi-spectral wide-swath camera, and takes into account both revisit cycle and resolution．BNU-1 is of great significance for the study of climate and environment change and monitoring in polar regions, at high coverage and medium resolution．However, the band noise of level 0 data masks the ground feature of image and does not reflect the real radiation characteristics of ground features, seriously affecting data application．Therefore, in this paper a method is proposed to remove the strip based on statistical principle for wide-swath image．The algorithm uses the dark noise extraction method corresponding to the image of 0° solar altitude angle and the uniform scene mask method to extract offset and gain values of each detector respectively, and realizes the strip removal．Experimental data indicate that compared to original data, this algorithm robustly removes band noise, restores real surface radiation information, and gives full play to the application value of multispectral data of BNU-1．