Correção atmosférica em imagens hiperespectrais : avaliação de diferentes metodologias em condições tropicais
There are many hyperspectral sensors available in the field, aerial, and orbital levels. Among the airborne sensors, the AisaFENIX hyperspectral sensor, used in this work, stands out. This sensor allows a data acquisition of terrestrial aircraft with high spectral detail, but it has no works in Braz...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis doctoral |
| Estado: | Versión publicada |
| Fecha de publicación: | 2017 |
| País: | Brasil |
| Institución: | Universidade Estadual de Maringá (UEM) |
| Repositorio: | Repositório Institucional da Universidade Estadual de Maringá (RI-UEM) |
| Idioma: | portugués |
| OAI Identifier: | oai:localhost:1/4635 |
| Acceso en línea: | http://repositorio.uem.br:8080/jspui/handle/1/4635 |
| Access Level: | acceso abierto |
| Palabra clave: | Sensor AisaFENIX Imagem hiperespectral Atenuação atmosférica Linho empírica Lonas coloridas Ciências Agrárias Agronomia |
| Sumario: | There are many hyperspectral sensors available in the field, aerial, and orbital levels. Among the airborne sensors, the AisaFENIX hyperspectral sensor, used in this work, stands out. This sensor allows a data acquisition of terrestrial aircraft with high spectral detail, but it has no works in Brazil evaluating a quality of data acquired through this sensor, making it essential, due to the Brazilian climatic conditions, that can affect the quality of the images. The objective of this work it is consist to perform the atmospheric correction of hyperspectral images acquired from the AisaFENIX sensor, using algorithms and known empirical methods, besides colored tarps arranged in the field at the time of image acquisition. In this work, were used tarps in the colors: blue, yellow, red, green, light gray and dark gray, and a BaSO4 plaque. The colored tarps demonstrate uniformity and do not suffer changes in relation to time, serving as control targets in the process of image correction. The algorithms used in this work were FLAASH, ATCOR-4, QUAC and the Empirical Line method (using the tarps in the calibration process and without using the tarps). Three images located in the UEM were selected and collected on April 30th, May 1st and 2nd, 2016, for to be corrected using the five methods previously mentioned. The ATCOR-4 and Empirical Line methods generate, from the use of the tarps and the BaSO4 plaque, calibration files that were used in the process of correcting other images. In this way, a fourth image, acquired on May 1, 2016, was used to evaluate if the calibration files generated from the ATCOR-4 and Empirical Line, improved the quality of the atmospheric corrections in images that did not contain targets for calibration. The spectral curves of the targets extracted from the images were compared with the spectral curves collected in the field with the ASD sensor. The Pearson correlation coefficient, determination coefficient, correlation test, mean comparison test, and SAM and ASDS index were used in the analysis of spectral curves. The correction methods that presented the best performances according to the SAM and ASDS indexes were ATCOR-4, QUAC and LE with tarps. The FLAASH did not generate satisfactory corrections in any of the images. The LE without tarps presented overestimation of the spectral values in relation to the field data. The yellow, red and green tarps proved to be suitable targets for calibration, in all images their correlations were higher than 0,8 in the visible and infrared regions. The blue tarp did not present satisfactory results, mainly in the infrared region, with correlations lower than 0,6. The ATCOR-4 calibration file used in image 4 processing improved the quality of the correction according to the SAM and ASDS indexes. The calibration file of the Empirical Line method did not contribute to the correction of the image, demonstrating not to be a proper methodology in the correction of images in different atmospheric conditions |
|---|