I am about to carry out research in an urban area to detect the change over a period using multi-temporal Landsat images. I am adopting post classification assessment technique. What are the necessary corrections required? With which one must I start? Thanks for cooperation.
Before starting any processing u have to do the geometric corrections
thanks Xiaoma Li and Sudhansu
my images are already geometrically corrected, what after that ?
Before doing any change detection you need to do the relative radiometric normalization (RRN) using pseudo invariant features(PIF's). This is easy in case of urban change detection studies. You need to normalize slave image wrt master image in case of bitemporal studies and in case of multitemporal keeping one image as master rest are to be corrected as slaves. Best of luck.
To get an accurate result by applying the change detection techniques using multi-temporal Landsat satellite images you should assemble images were acquired in the same week of the month or at least in the same month. For example July of 2012, 2003, and 1996. Then you need to do the image registration, I mean image to image registration. You can adopt the post classification comparison or the image differencing to get the spatial and temporal changes. Good luck
- Badges
- Science topic
- Similar topics
- Psychology