To date, several change detection methods have been developed using
conventional image differencing (consider change in reflection), using image ratio,
normalized difference vegetation index, principal component analysis, multi-date
image classification, post-classification comparison, manual onscreen digitization
(Jensen 1986, Jensen et al. 1993, Mas 1997, El-Raey et al. 1999). In this study, a
series of methods were employed to identify land use changes in the coastal areas of
Ban Don Bay, Thailand. The most common approach is digitizing and overlaying to
compare the data sets from two dates in order to detect land use changes of the first date data that have been replaced by the new land use in the second date. The
second method is image differencing, which simply compares the categories between
images. Another method, NDVI differencing, is used to compute different NDVI
values between two dates to compare vegetation characteristics; and the last
method, NDVI composite (Sader and Winne 1992), is a method that uses three-date
NVDI values to detect activities and change or no change during a three-date
period.