by factor analysis are intended to make it possible for the user to identify 

 physical phenomena responsible for variation in the data to be identified 

 with particular factors. Upon completion of this statistical analysis the 

 processing coefficients thus specified were input to the analog processing 

 unit, and film showing the variation in the first three factors was produced. 

 The results of this factor analysis are shown in Figure 2. The ground 

 truth collected by Purdue University identifies the dark areas in the 

 factor 1 imagery as corn fields. In factor 2 the dark areas correspond to 

 fields of bare soil. The second statistical analysis employed with respect 

 to this data was multiple linear regression analysis. Samples of this 

 imagery are shown in Figure 3. The middle strip identified at the top with 

 the caption "corn equals white" corresponds rather well with the factor 1 

 imagery from figure 2. Corn is enhanced in both cases. The sense of the 

 enhancement is changed however so that in the case of regression, corn 

 appears as the lighter color fields. With the aid of the ground truth, wheat 

 fields were also identified. The regression analysis was performed to 

 determine linear combinations which would enhance wheat fields. This is 

 also identified in Figure 3. Finally, bare soil was enhanced in much the 

 same manner as found in Figure 2. Again, however, the sense of the 

 enhancement is reversed so that bare soil appears as the light color in 

 the regression imagery while it appears as a dark color in the factor 



27-5 



