A new method of restoring ETM+ SLC-off images based on multi-temporal images

Regular gaps present on the SLC-off scenes and about 25% data is lost since the failure of the scan line corrector (SLC) in 2003. The data itself preserves good radicalization and geometry performances, and gaps occurred randomly in different images, so it's possible for cross restoration. A new method to restore the SLC-off data using another ETM+ image with near time and same extent is proposed. The first step is to eliminate the affection of “foreign difference” (atmospheric condition, weather, etc) of the two images (image needs restored and image for filling). Transform the image for filling by local relative radiometric normalization. Then convert the image for filling according to nearby spectral-similar pixels by building up a weight that considers both spectral and spatial distance. Experiment result indicates that the method is an effective method for recover ETM+ images and is superior to traditional local regression match approach.