Geo-location correction of CBERS 2B imagery using fast template matching on a GPU

Abstract

CBERS 2B satellite imagery was acquired in the period 2007-2010, at 2.5 m (panchromatic) and 20 m (multispectral) spatial resolution, respectively. Much of the archived data is available, free of charge, via the Brazilian space organization INPE. However, to be suitable for wide area mapping at scales 1:50,000 to 1:100,000 scales (or better), errors in geo-location, which range from several 100s of meters to several kilometers, must be corrected for each individual image frame. We demonstrate a novel method which deploys very fast template matching using GPU accelerated convolution. The method can be used with template samples from a globally consistent reference set. We demonstrate this with the use of the Terracolor LANDSAT product over Brazil and with Bing maps extracts for some urban areas in panchromatic scenes. The main advantage of the method is the possibility to use a large number of templates of a significant size (256 by 256, or 512 by 512 pixels). Processing time remains below 1 minutes per frame, which includes template selection via a WMS service. The result of the match is supplied as a list of ground control points, which we store in the VRT format, allowing the use of GDAL for transformation. We discuss implementation details and show how we can apply the method to other sensor combinations as well.

Geo-Correction of High-Resolution Imagery Using Fast Template Matching on a GPU in Emergency Mapping Contexts

“…Even though this is only 6-times the nominal, at-nadir, image pixel spacing of 0.50 m, correction with ground control points is required to bring this accuracy within a 1 pixel accuracy, which would be required for consistent mapping at scales 1:2,000-1:5,000, which is feasible with this image quality. Legacy satellite sensors have positional accuracies in the 100-1,000 m range, but a number of operational and historical sensors exist for which positional accuracy can vary significantly [8].…”

“…In [8], we presented a novel approach using very fast large template matching applied to full resolution images and large templates. This approach was developed on the basis of previous work [15] and deploys the phase correlation image alignment method, which is based on the Fourier Shift Theorem (see [10] for a comprehensive review).…”

“…In case the native geolocation accuracy of the image is variable and of low quality [8], resampling the source image to a lower resolution may be considered. The correction resulting from the template matching at this lower resolution can then be applied to the full resolution source image, after which a full resolution match can be applied to find the final correction.…”