This work presents an implementation of the fast Haar wavelet transform (FHWT) mathematical algorithm assisted by the Mallat algorithm together with the ARSIS concept so as to successfully attain satellite image merging with different resolutions. Four pairs of images, namely (multispectral—pancromatic): IKONOS, GeoEye, OrbView-2, and Landsat ETM+, representing different environments are used to assess the present implementation of FHWT. In order to compare the performance of FHWT with other orthogonal and bi-orthogonal wavelets, the same satellite images are merged using other five wavelets from some of the MATLAB’s available sets, namely bior6.8, rbio6.8, db7, dmey, and Haar. After applying the six wavelets to the four regions under study, four indices are used to assess spatial and spectral quality of the merged images, namely correlation coefficient, relative average spectral error, relative dimensionless global error in synthesis, and the universal quality index Q. Moreover, shape recognition capacity is also assessed based on the resulting merged images. To do so, various objects are binarized in each of the images. Binarized versions of the objects are compared to the same objects obtained from their corresponding panchromatic image. These binarized versions are also assessed using kappa coefficient and overall accuracy. For each of the indices, the best results are obtained with the proposed method FHWT.