Spectral and spatial assessment of the TDW Wavelet transform decimated and not decimated for the fusion of OrbView-2 satellite images

Abstract

The objective of this article is to develop and evaluate two methodologies that allow to improve the spatial resolution without significant loss of the spectral resolution of a multispectral image (MULT) and panchromatic (PAN) OrbView-2. In the first method, the algorithm is used: Discrete Transformation of Decimal Wavelet (TWD decimated) or Mallat Algorithm. The Value is obtained from the MULT. Then applying the fusion to the Value and PAN component through the TWD decimated daubechies (db4) generates a new Value (nval-m). With the nuance and saturation of the MULT image, the inverse HSV-RGB transformation is performed to generate a new multispectral image (N-MULT1). In the second methodology, the algorithm is used: Discrete Wavelet Transform not decimated (TWD not decimated) or Algorithm of À trous, applying the fusion to the Value component and PAN through the TWD not decimated generates a new Value (nval-a) and with the nuance and saturation of the MULT image the inverse HSV-RGB transform is made to generate a new multispectral image (N-MULT2). Finally, the results of the two methods are presented using the ERGAS, RASE and Qu indices for assessment. It is obtained that the à trous method is better spatially and spectrally.