Descripción: |
The performance of multiplicative and Intensity Hue Saturation (IHS) techniques to fuse RADARSAT-2 with IKONOS imagery for land cover mapping was assessed. Four major land covers were identified in each case: water, inland forest, wetland forest and urban areas. The assessment of the classification accuracy shows that a higher overall accuracy of 83.33% was achieved with the IHS fused image (kappa coefficient of 75.61% against an overall accuracy of 75.00%). The multiplicative technique reached a kappa coefficient of 63.83%. Five statistical fusion accuracy assessment indicators were used to assess the performance of the fused techniques. They are: mean (μ), standard deviation (σ), Relative Difference Mean (RDM), Relative Variation Difference (RVD) and Universal Image Quality Index (Q). The results indicate that the HIS technique has a better performance (μ=93.829, σ=18.338, RDM =0.5487, RVD = -0.8768, Q=0.0184) compared to the multiplicative fusion technique (μ=21.703, σ=21.773, RDM = -0.6417, RVD = -0.8263, Q=0.0134). It is therefore concluded that IHS fusion technique is more appropriate in preserving the spectral information of the multispectral image for the mentioned application. |