Change detection studies in remote sensing operate with the notion that a quantifiable difference in an object’s spectral value, from one time period to another, represents a physical change on the ground. To confound this premise, other factors, such as atmospheric conditions and illumination geometry, can influence an object’s spectral response. For this reason, a common first step in digital change detection is the task of image-to-image normalization. In this Technical Note, we present an efficient method for radiometric normalization of images by converting Landsat Thematic Mapper (TM) and Enhanced Thematic Mapper (ETM) pixel values into their respective ordinal ranks. To demonstrate this normalization approach, raw and ranked Landsat near-infrared (NIR) image pairs, with a 6-year lag, were differenced to detect change in forest cover located in central British Columbia, Canada. Results demonstrate that ranking values prior to image differencing improves detection of change. The ease and efficiency of the approach is promising for automation and studies of change over large areas.
Mendeley saves you time finding and organizing research
Choose a citation style from the tabs below