Comparing pan-sharpening algorithms to access an agriculture area: a mississippi case study

Authors: Fletcher, Reginald

Submitted to: Agricultural Sciences
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/1/2023
Publication Date: 9/3/2023
Citation: Fletcher, R.S. 2023. Comparing Pan-Sharpening Algorithms to Access an Agriculture Area: A Mississippi Case Study. Agricultural Sciences. Agricultural Sciences, 14, 1206-1221. https://doi.org/10.4236/as.2023.149081.
DOI: https://doi.org/10.4236/as.2023.149081

Interpretive Summary: The fusion of high-resolution panchromatic imagery with lower resolution multispectral satellite imagery has shown promise for monitoring natural resources and farming areas. In rural areas of Mississippi, agricultural landscapes can range from complex mixtures of vegetation and built-up areas to dense vegetative regions. These areas are ideal for satellite remote sensing and information is lacking on using pan-sharpened satellite to survey agricultural landscape in rural areas. A scientist at the USDA-ARS, Crop Production Systems Research Unit, Stoneville, MS used free and open-source software to test and identify image fusion techniques that were useful in merging panchromatic and multispectral satellite to assess agricultural landscape consisting of a combination of landcover features found in rural areas of Mississippi. The research was completed with open-source software (PanFusion and ImageAnal) available to the public. These findings should be useful to industry, government, and the public.

Technical Abstract: Numerous satellites collect imagery of the earth's surface daily, providing a wealth of information to the public and private sectors. For years, coarse, medium, and high spatial resolution satellite systems have provided the necessary data to access agricultural regions. The fusion of high-resolution panchromatic imagery with lower resolution multispectral satellite imagery has shown promise for monitoring natural resources and farming areas. In rural areas in Mississippi, agricultural landscapes can range from complex mixtures of vegetation and built-up areas to dense vegetative regions. This study was conducted to assess the influences of farming landscapes on the image quality of pan-sharpened imagery. WorldView 3 satellite imagery of an area consisting of landscape commonly found in rural areas in Mississippi was subjected to 17 pan-sharpening algorithms. Eight quality indices were used to compare the images. The best overall fusion algorithm was a trous wavelet transform with the injection model 3 (ATWT_M3). Brovey transform and intensity, hue, and saturation fusion methods were ranked in the middle and as low fusion algorithms by the indices. Those fusion algorithms often compromised the color of the landscapes. The smoothing filter-based intensity modulation algorithm added artifacts to the images. Pansharpened satellite imagery has great potential as a tool for enhancing survey of agricultural areas. The key to success is selecting an image fusion process that does not compromise the image integrity. The PanFusion open-source free tool used in this study was valuable in determining the appropriate fusion process.