Three techniques are described for the extraction of surface emissivity and compositional information from Thermal Emission Imaging System (THEMIS) images. Synthetic images were constructed with different atmospheric properties and random and systematic errors to estimate the uncertainty in the retrieved surface emissivity. The three techniques are as follows: (1) A constant radiance removal algorithm determines and removes a constant radiance from atmospheric emission as well as systematic calibration radiance offsets. (2) Surface emissivity retrieval uses low-resolution Thermal Emission Spectrometer surface emissivity data to determine atmospheric properties for an image that can be then applied to individual THEMIS pixels. (3) A spectral unit mapping algorithm determines spectral end-member concentrations using a deconvolution routine similar to several previous applications. The initial application of these techniques to three images covering the same region of the Martian surface but at different surface temperature and atmospheric conditions yields consistent surface spectral shapes as well as end-member concentrations. Retrieved aerosol opacity information is consistent with an independently developed opacity retrieval method. The application of the techniques described here is done in a stepwise fashion and may be applied to the desired level of analysis necessary for interpretation of surface properties.