Pan-sharpening & image fusion

Title of the bachelor thesis: Pan-sharpening, Image Fusion – comparison common used algoritms

The thesis was elaborated by Michaela Čermáková, student of Geography-geoinformatics, Faculty of Natural Sciences, Palacky University in Olomouc. The whole thesis is written with the aim of showing a summary of algoritms, which can be used for pan-sharpening. This topic deals with Remote Sensing. Pan-sharpening combines the colour information from a multi-spectral file, with the geometric information from the panchromatic band.

The work includes a theoretic section of algoritmic description, which are principle component analysis (PCA), intensity-hue-saturation (IHS), Brovey transform, high-pass-filtering (HPF), muliplicative, Ehlers fusion and wavelet transform. In the apllied section were these algoritms tested out on data from sattelite LANDSAT. Algoritms were tested in ERDAS IMAGINE 9.3 and extensions Image Analyst for ArcGIS 9.3.

Comparative evaluations of the different approaches have examined different combinations of fused data using different criteria, so it is difficult to derive definitive report. PCA and IHS methods largely disregard important spectral characteristics of specific satellite sensors, therefore no consistent, colour-preserving results can be achieved. There is limit to the number of bands that can be processed at one time using IHS or Brovey transform. So it is better to use HPF or Ehlers fusion with implemented Fourier analysis, which removes ringing effect and stripping.