Implementation and evaluation of the High Pass Filter and Wavelet À Trous transforms in Matlab to fusion Landsat 8 OLI/TIRS Satellite Images

Abstract

In this article, the High Pass Filter-HPF, and Wavelet À trous transforms are developed mathematically, to later implement them in Matlab. The fusion of satellite images is performed with each one of the implemented transforms, with a proposed methodology. A Landsat 8 OLI/TIRS image (Panchromatic - PAN and multispectral - MULTI) of a northwestern sector of Peru -where sugar cane crops are evident- is used to generate two fused images, namely: MULTIHPF and MULTITWA. The fused images were evaluated both in spatial and spectral quality through four indices, specifically: correlation index, ERGAS, RASE and Q index, in order to determine the efficiency of the proposed methods. Best results of the spectral evaluation were obtained with the MULTITWA image, achieving correlations higher than 0.95, a Q index of 0.96 and a RASE value of 9.2%, while spatially higher values than 0.96. Regarding spatial richness, the best results were obtained with MULTIHPF with an ERGAS of 14.6, a RASE of 29.3% and a Q of 0.7.