Infrared imaging is commonly used for performing thermography based on field calibration that simply relates image levels to apparent temperature levels using field blackbodies. Under normal conditions, the correlation between the image levels and blackbody temperature is strong, allowing conversion of the raw data into units of blackbody-equivalent temperature without consideration of other factors. However, if certain instrument anomalies are present, a compensation procedure that involves more in-depth sensor characterization may be required. The procedure, which uses an analysis of temperature-dependent dark current, optical emissions, and detector response, is described along with results for a specific case. The procedure involves first cold soaking a thermal camera and then observing the cooldown behavior of the sensor under non-stressing conditions. Variations in environmental temperature levels are then used to observe cooler performance and dark current levels. A multi-variate linear regression is performed that allows temperature-dependent dark current, lens emission, lens transmission, and detector quantum efficiency to be fully characterized. The resulting data describe for each image pixel a relationship between the scene temperature and the observed values of image signal, detector temperature, and camera temperature. The procedure has been applied successfully to a thermal imager used to collect field data while suffering from instrument anomalies due to a faulty cooler. Using the resulting characterization data for the pixel-dependent dark current, image data collected with the thermal imager was compensated. The compensation involved using spatial filtering to determine temperature shifts caused by the faulty cooler based on the predictable pattern of pixel-to-pixel variations in dark current. The estimated temperature shift was used to compute a compensation offset for each pixel based on its known dark current coefficient. The compensated image data, while still degraded, was sufficiently corrected for the predictable effects of dark current variations to allow valid thermography to be performed.