The levels of noise in Hyperspectral data are quite different from band to band. Junk band refers to the band which is so noisy that it is usually discarded in data analysis. Considering that the profiles of bands at close wavelengths are quite similar and the conlourlet is good at capturing profiles, we propose a junk band recovery algorithm for hyperspectral data based on contourlet transform. Both the noisy bands and the noise free bands are transformed by contourlet band by band. The high frequency coefficients in junk bands are replaced with weighed sum of the high frequency coefficients in noise free bands, and the low frequency coefficients remain the same to keep the main spectral characteristics from being distorted. Junk bands then are recovered after inverse contourlet transform. The performance of our method is tested on the hyperspectral data cube obtained by Operational Modular Imaging Spectrometer (OMIS). Experimental results show that the proposed method is superior to the traditional denoising method BayesShrink in both computing time and peak-signal-to-noise ratio (PSNR) of recovered bands.