One of the initial steps in the preprocessing of remote sensing data is the atmospheric correction of the at-sensor radiance images, i.e., radiances recorded at the sensor aperture. Apart from the accuracy in the estimation of the concentrations of the main atmospheric species, the retrieved surface reflectance is also influenced by the spectral calibration of the sensor, especially in those wavelengths mostly affected by gaseous absorptions. In particular, errors in the surface reflectance appear when a systematic shift in the nominal channel positions occurs. A method to assess the spectral calibration of hyperspectral imaging spectrometers from the acquired imagery is presented in this paper. The fundamental basis of the method is the calculation of the value of the spectral shift that minimizes the error in the estimates of surface reflectance. This is performed by an optimization procedure that minimizes the deviation between a surface reflectance spectrum and a smoothed one resulting from the application of a low-pass filter. A sensitivity analysis was performed using synthetic data generated with the MODTRAN4 radiative transfer code for several values of the spectral shift and the water vapor column content. The error detected in the retrieval is less than +/- 0.2 nm for spectral shifts smaller than 2 nm, and less than +/- 1.0 nm for extreme spectral shifts of 5 nm. A low sensitivity to uncertainties in the estimation of water vapor content was found, which reinforces the robustness of the algorithm. The method was successfully applied to data acquired by different hyperspectral sensors.
In this paper, we propose a procedure to reduce dimensionality of hyperspectral data while preserving relevant information for posterior crop cover classification. One of the main problems with hyperspectral image processing is the huge amount of data involved. In addition, pattern recognition methods are sensitive to problems associated to high dimensionality feature spaces (referred to as Hughes phenomenon or curse of dimensionality). We propose a dimensionality reduction strategy that eliminates redundant information by means of local correlation criterion between contiguous spectral bands; and a subsequent selection of the most discriminative features based on a Sequential Float Feature Selection algorithm. This method is tested with a crop cover recognition application of six hyperspectral images from the same area acquired with the 128-bands HyMap spectrometer during the DAISEX99 campaign. In the experiments, we analyze the dependence on the dimension and employed metrics. The results obtained using Gaussian Maximum Likelihood improve the classification accuracy and confirm the validity of the proposed approach. Finally, we analyze the selected bands of the input space in order to gain knowledge on the problem and to give a physical interpretation of the results.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.