Sorting fused images for multi-time analysis of the area surrounding the headwaters of the meta river

Abstract

This paper focuses on the process of theme-sorting Landsat images that have been enhanced by means of multispectral-panchromatic fusion. In addition to the assessment of the fusion methodologies, the paper also highlights the changes that have occurred in the area surrounding the headwaters of the Meta River during the last 16 years, near the municipality of Puerto López (Meta – Colombia). To carry out the fusion process, the Wavelet transform was used. The transform captures suitable information about spatial details of a panchromatic image and integrates the resulting image into the multispectral bands. Decision trees were used to classify the set of fused images. Classification with decision trees was based on differential discrimination of the spectral ranges for each of the coverage areas associated to multi-spectral bands. The present study delivers five theme maps showing the changes that have been occurred in related areas which reach a precision in the classification of 93.01% for the fused image in comparison of 74.08% obtained for the image without fusion.