Novel method of satellite image correction for multi-temporal analysis of environmental changes purposes
Abstract
Multi-temporal, long-term environmental analyzes require imagery that are diverse in terms of spatial, radiometric and spectral resolution. Satellite imagery acquired with various sensors over more than 30 years are usually used for such analysis. To carry out analyzes it is necessary to perform the correction of archival images. The classical radiometric correction (including solar, topographic and atmospheric correction) is not sufficient for correct analysis of such data set. The classic radiometric correlation is only the initial stage of the total correction of the set. In order for the results of multi-temporal analyzes to be reliable and more precise, it is necessary to bring the data to the same spatial, spectral and radiometric resolutions.Resampling the spatial resolution in the context of multi-temporal analyzes of the environment is not the only process. Except that also the extraction of characteristic details to further analysis mandatory. The publication presents a new method of image resampling based on wavelet analysis. Resampling will standardize the resolution of the spatial set. In addition, spectral image resolution also plays an important role in the detection of environmental changes. It is crucial that the subsequent images have a similar spectral resolution.A new spectral resolution correction algorithm based on sequential fusion of images acquired in subsequent dates was proposed, where the resulting images are a weighted average of the data to be integrated. The weights depend on the content of spatial and spectral information in the processed images. As part of the research experiment, new corrected spectral channels of images were determined, the same for all images in a given integration sequence. The similarity of corrected images to original images is over 60%, which guarantees that important spectral information recorded in each subsequent image is preserved.
- Publication:
-
42nd COSPAR Scientific Assembly
- Pub Date:
- July 2018
- Bibcode:
- 2018cosp...42E1727K