Advanced calibration of fourier-transform spectrometers

遥感应用需要尽可能多的信息，以便最大限度地确定场景特征。为了做到这一点，基本方法包括图像处理（给出二维空间细节）和光谱处理（给出光谱细节，光谱细节是能量在许多不同的“颜色”上的重新分配，对应于窄的光学波段）。现在的仪器是为了同时提供空间和光谱信息而开发的。它们被称为高光谱成像仪或成像光谱仪。这些仪器提供每个像素的光谱，反过来为每个光谱带提供图像，从而产生三维数据立方体（最终代表大量数据）。 理想的数据立方体将仅包含场景信息，而不包含任何仪器伪影。然而，仪器的贡献总是合并与场景特征的测量数据。因此，必须开发尽可能完全地去除数据立方体中每个点的仪器贡献的方法。这个过程称为校准。正确校准大型数据立方体是一项艰巨的任务。 提出的研究的目的是研究一种特殊的成像光谱仪，傅里叶变换光谱仪。我们计划通过使用仪器光谱响应函数来精确预测其行为，该函数与其增益和线形有关。我们接下来将开发准确的方法来校准成像傅里叶变换光谱仪的增益和线形状的知识的基础上产生的数据。实验室成像傅里叶变换光谱仪将用于测试理论预测。在我们的研究计划中，我们还将开发数字有效的方法来实现所提出的校准算法。 因此，我们的研究面向实现更有效和可靠的仪器，这要归功于对它们基本行为的更好理解。这种优化的仪器成为关键的地面，飞行和空间项目的关键组成部分，旨在民用和国防应用。因此，我们的目标是开发新的仪器，以更低的成本/质量/体积显示出更好的性能。

Remote sensing applications require as much information as possible in order to maximize the identification of scene characteristics. In order to do so, basic approaches include processing of images (giving spatial details in two dimensions) and spectra (giving spectral details, which are the repartition of the energy across many distinct "colours", corresponding to narrow optical bands). Nowadays instruments are developed to provide simultaneously spatial and spectral information. They are called hyperspectral imagers or imaging spectrometers. Such instruments provide spectra for each pixel, conversely images for each spectral band, resulting therefore in three-dimensional data cubes (representing eventually huge amount of data).      Ideal data cubes would contain only scene information, without any instrument artefacts. However, instrument contributions are always merged with scene characteristics in the measured data. One has thus to develop approaches to remove, as completely as possible, the instrument contributions for every point in the data cube. This process is called calibration. It is a tremendous task to properly calibrate large data cubes.      The aim of the proposed research is to study a particular imaging spectrometer, the Fourier-transform spectrometer. We plan to precisely predict its behaviour by using the instrument spectral response function, which is related to its gain and its line shape. We will next develop accurate methods to calibrate data produced by imaging Fourier-transform spectrometers based on gain and line shape knowledge. Laboratory imaging Fourier-transform spectrometers will be used to test theoretical predictions. Within our research program, we will also develop numerically efficient ways to implement the proposed calibration algorithms.      Our studies are thus oriented towards the realisation of more efficient and reliable instruments thanks to better understanding of their fundamental behaviour. Such optimised instruments become the key component in critical ground, flight and space projects, aiming for civil as well as defence applications. Our goal is therefore to develop novel instruments showing improved performances for lower cost / mass / volume.