410 
image ratioing (Weismiller and Kristoff, 1977). Several other tech 
niques have been developed using multivariate statistical methods, but 
these have not been fully explored. In particular, factor analytic (FA) 
techniques have been proposed but not fully investigated (Byrne et al., 
1980). Using digitized aerial photos supplied by the Defense Mapping 
Agency (DMA), under contract No. DMA-800-81-C-0031, a study has been 
conducted comparing several existing algorithms for analyzing temporal 
information to a FA technique. 
2. OBJECTIVES 
The primary objective of this study was to evaluate the use of digital 
processing techniques for detecting temporal change in land use/land 
cover from digitized aerial photographs. Using existing software capa 
bilities, two aerial photos were analyzed and compared to highlight 
areas of land use change. Four techniques were evaluated on an initial 
test site and the best method was applied to four additional test sites. 
3. STUDY AREAS 
For each site, an early photograph (the reference image) and a later 
photograph (the mission image) were supplied. Each site had several 
land-cover features that had changed between the time of exposure of 
the two photos. The test-site areas on each photo were digitized with 
a microdensitometer, at a resolution of 20 pixels per millimeter, and 
the responses recorded on magnetic tape. These data were then refor 
matted, geometrically corrected, analyzed for change detection, and 
evaluated for correct analysis. 
Site 3, selected as the initial training site, was a sewage treatment 
facility near Fort Belvoir, Virginia. The data for this site consisted 
of one October 1972 black-and-white (BW) photo, digitized on one channel 
so that low values represented white (low photographic density) and high 
values represented black (high photographic density); and one April 1978 
color photo, digitized onto three channels (red, blue, and green) so 
that low values represented low intensity hues and high values repre 
sented high intensity hues. The nominal scale of the BW photo was 
1:10,000 while that of the color photo was 1:12,000. 
Test sites 2 and 5 were from the same flights as test site 3; therefore, 
the same data formats were supplied. Test site 2 was a residential area 
at Fort Belvoir and test site 5 was a rural-residential area near 
Gunston Hall Plantation, south of Fort Belvoir. 
Data for test site 4 consisted of one 1:48,000 March 1968 BW photo and 
one 1:53,000 November 1975 color photo digitized into three channels. 
This site was a rural-residential area west of Fort Belvoir. 
Data for site 7 consisted of two BW photos, from January 1953 and from 
February 1978, both at the scale of 1:30,000. This site was an airfield 
on North Island, near San Diego, California. 
4. PROCEDURES 
Initially, the data set for each test site was reformatted for compati 
bility with the ORSER software system (Turner et al., 1982). The data 
sets were then displayed and examined for differences in registration 
between the early and late data. The early and late data were then 
geometrically corrected to each other to account for these differences.