The retrieval of surface reflectance information from the same single pixel of the Hyperion image atmospherically corrected by using image-based [internal average relative reflectance (IARR), log residuals, and flat field] and radiative transfer model (RTM)-based [the fast line-of-sight atmospheric analysis of spectral hypercubes (FLAASH) and the Atmospheric and Topographic Correction 2 (ATCOR-2)] approaches and the spectral feature characteristics of this information were quantitatively and comparatively examined based on measured ground spectral reflectance data. The spectral features quantitative analysis results of the reflectance data showed that spectral reflectances that are suitable and best fitting to the ground spectral reflectances which were obtained from the pixels of FLAASH, ATCOR-2, and flat field–corrected images, respectively. The retrieval of surface reflectance from the FLAASH-corrected image pixels, in general, produced high scores in spectral parameter analyses. Of the image-based approaches, only in flat field–derived reflectance data, results were obtained which are high and nearest to those of RTM and ground spectral reflectance data. Generally, low scores obtained in the spectral parameter analyses of the surface reflectance values retrieved from single pixels of IARR and log residuals-corrected images showed the results that fit worst to the measured ground spectral reflectance.