| Image data obtained from pushbroom sensors such as the Advanced Land Imager (ALI) contains striping artifacts due to mismatches in the response between individual detectors in the array. Characterization and removal of these artifacts is called relative gain correction. Histogram equalization, which uses the first and the second order statistics for each detector, is a common technique used to remove such striping. This approach is valid in the case of whiskbroom sensors where each detector sees the same image information in a statistical sense; however, it is not applicable to pushbroom sensors where each detector does not essentially see the same image information. If however, the range of data is extended over many scenes, this assumption can be used and relative gain can be estimated. Based on this concept, a relative gain characterization algorithm has been developed and implemented which produces corrections equal to or surpassing corrections based on pre-launch estimates of relative gains. |
