Radiometric correction for linear change-detection techniques: analysis in bi-temporal space

Radiometric corrections can remove or reduce inconsistencies induced by temporal factors in multi-temporal images, but their effectiveness for change detection is still unclear. This study applied a physics-based model to pseudo-invariant features (PIFs) and showed that absolute correction is in principle equivalent to relative correction in change-detection studies if the spatial heterogeneity of the atmosphere is not rigorously considered. Bi-temporal space analysis revealed that the performance of a linear change-detection algorithm depends on whether the determined slope of the no-change zone coincides with that of the PIF line, offering insight into the effectiveness of radiometric correction. A case examination showed that either absolute or relative correction adjusted the paired image bands to a common standard. However, the corrections did not necessarily reduce the bias in the zone slope determined by the change-detection algorithm, illustrating the futility of performing either absolute or relative corrections for linear change-detection techniques, except for image differencing.