University of Twente Student Theses
Image Pan-sharpening with Markov random field and simulated annealing
Kitaw, Hailemariam Gedlu (2007) Image Pan-sharpening with Markov random field and simulated annealing.
|
PDF
5MB |
| Abstract: | As many earth observation satellites provide both high resolution panchromatic and low resolution multispectral images, pan-sharpening is an important technique in the field of remote sensing to produce a high resolution multispectral image for a given low resolution multispectral image and a higher resolution panchromatic image of the same area. To date, several image pan-sharpening techniques like the IHS method, the PCA method, the wavelet method, etc have been developed. However for the recent high and very high resolution satellite images the developed algorithms can hardly produce satisfactory results [44, 45]. There are significant color distortions in the pan-sharpened image and the quality of these pan-sharpened images is also seldom questioned. This research concentrates on developing a new method for image pan-sharpening in Bayesian image restoration framework using Markov Random Field (MRF) and simulated annealing that provide maximum a posteriori (MAP) estimate of the pan-sharpened image. The study models the (unknown) pan-sharpened image with an MRF (prior probability) and models observation process for the panchromatic and the coarse resolution multispectral images (conditional probability). These probabilities are combined to derive posterior probability for the pan-sharpened image. Optimal solution is sought as the image that maximizes the posterior probability. The simulated annealing algorithm is used to obtain the optimal solution. In order to test the performance of the method, a reference image is chosen and is assumed as the unknown pan-sharpened image. The reference image is degraded to produce panchromatic and multispectral images. The method is then applied to restore the pan-sharpened image from these degraded images and the accuracy assessment is performed by comparing it with the reference image. The method shows a very high correlation between the restored pan-sharpened image and the reference image. The performance of the proposed method is compared with other pan-sharpening methods and is proved that it performs better than existing methods. |
| Item Type: | Essay (Master) |
| Faculty: | ITC: Faculty of Geo-information Science and Earth Observation |
| Programme: | Geoinformation Science and Earth Observation MSc (75014) |
| Link to this item: | https://purl.utwente.nl/essays/105272 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page