Multisensor data fusion has been widely used in response to complementary nature of many data sets. This paper compares the results of four different data fusion methods used to merge Landsat-7 ETM+ and RADARSAT-1 Wide 1 data. The comparison was based on spectral characteristics of images using statistical and visual analyses of generated products. Four methods were used in the Landsat-7 ETM+ and RADARSAT-1 W1 data fusion: i) The best three bands combination(Landsat-7) based OIF (Optimum Index Factor) selection were merged with RADARSAT-1 data; ii) Decorrelation stretch was applied in the three bands (Landsat-7) selected by OIF and merged with RADARSAT-1 image; iii) PCA (Principal Component Analysis) to six reflective ETM+ bands (1, 2, 3, 4, 5 and 7) and posterior fusion of the three first Principal Components (PC1, PC2, PC3) with SAR; iv) A new approach SPC-SAR (Selective Principal Component - Synthetic Aperture Radar). The SPC-SAR product presented the best performance in the identification of coastal features and allowed the most effective enhancement of the different environments.
multisensor remote sensing; image fusion; Amazon Coastal Zone
