This thesis evaluated the effectiveness of sensor fusion--combining infrared and low-light-level image to improve the F/A-18 target standoff range requirement. Several human performance studies have shown inconsistent results regarding the benefits of color-fusion imagery. One method to test the validity of sensor fusion is to use mathematical models that simulate and predict the detection abilities of the human visual system. The mathematical models are derived quantitatively from the image statistics, while the behavior data are a qualitative measure of a human observer. This thesis developed a statistical analysis to compare and contrast these techniques to assess sensor fusion. The four models evaluated were: a Global matched filters, a Local matched filter, a Template matching filter, and a contrast-base image quality metric. Of the four models, the Global matched filter produced the highest degree of correlation with the human data. The Global matched filter moderately predicted which of the single-band sensors and which of the fined sensors provided the higher sensitivities despite the characteristically different scenes. Although there are many refinements that need to be explored, the Global matched filter concept may be used to evaluate and compare the many different fusion algorithms being proposed.