Fourier transform imaging spectrometers (FTISs) are widely used in global hyperspectral remote sensing due to the advantages of high stability, high throughput, and high spectral resolution. Spectrum reconstruction (SpecR) is a classic problem of FTISs determining the acquired data quality and application potential. However, the state-of-the-art SpecR algorithms were restricted by the length of maximum optical path difference (MOPD) of FTISs and apodization processing, resulting in a decrease in spectral resolution; thus, the applications of FTISs were limited. In this study, a deep learning SpecR method, which directly learned an end-to-end mapping between the interference/spectrum information with limited MOPD and without apodization processing, was proposed. The mapping was represented as a fully connected U-Net (FCUN) that takes the interference fringes as the input and outputs the highly precise spectral curves. We trained the proposed FCUN model using the real spectra and simulated pulse spectra, as well as the corresponding simulated interference curves, and achieved good results. Additionally, the performance of the proposed FCUN on real interference and spectral datasets was explored. The FCUN could obtain similar spectral values compared with the state-of-the-art fast Fourier transform (FFT)-based method with only 150 and 200 points in the interferograms. The proposed method could be able to enhance the resolution of the reconstructed spectra in the case of insufficient MOPD. Moreover, the FCUN performed well in visual quality using noisy interferograms and gained nearly 70% to 80% relative improvement over FFT for the coefficient of mean relative error (MRE). All the results based on simulated and real satellite datasets showed that the reconstructed spectra of the FCUN were more consistent with the ideal spectrum compared with that of the traditional method, with higher PSNR and lower values of spectral angle (SA) and relative spectral quadratic error (RQE).

Previous studies on the polarization imaging of underwater targets mainly focused on top-down detection; however, the capacities of bottom-up detection were poorly known. Based on in situ experiments, the capability of bottom-up detection of underwater targets using polarization imaging was investigated. First, to realize the objective of bottom-up polarization imaging, a SALSA polarization camera was integrated into our Underwater Polarization Imaging System (UPIS), which was integrated with an attitude sensor. At Qiandao Lake, where the water is relatively clear, experiments were conducted to examine the capacity of the UPIS to detect objects from the bottom up. Simultaneously, entropy, clarity, and contrast were adopted to compare the imaging performance with different radiation parameters. The results show that among all the used imaging parameters, the angle of polarization is the optimal parameter for bottom-up detection of underwater targets based on polarization imaging, which may result from the different diffused reflectance of the target surface to the linear polarization components of the Stokes vector.

Hyperspectral image classification is widely applied in several fields. Since existing datasets focus on a single scene, current deep learning-based methods typically divide patches randomly on the same image as training and testing samples. This can result in similar spatial distributions of samples, which may incline the network to learn specific spatial distributions in pursuit of falsely high accuracy. In addition, the large variation between single-scene datasets has led to research in cross-scene hyperspectral classification, focusing on domain adaptation and domain generalization while neglecting the exploration of the generalizability of models to specific variables. This paper proposes two approaches to address these issues. The first approach is to train the model on the original image and then test it on the rotated dataset to simulate cross-scene evaluation. The second approach is constructing a new cross-scene dataset for spatial distribution variations, named GF14-C17&C16, to avoid the problems arising from the existing single-scene datasets. The image conditions in this dataset are basically the same, and only the land cover distribution is different. In response to the spatial distribution variations, this paper proposes a constrained spectral attention mechanism and a constrained spatial attention mechanism to limit the fitting of the model to specific feature distributions. Based on these, this paper also constructs a constrained spectral–spatial attention residual network (CSSARN). Extensive experimental results on two public hyperspectral datasets and the GF14-C17&C16 dataset have demonstrated that CSSARN is more effective than other methods in extracting cross-scene spectral and spatial features.

Due to the development of spectral remote sensing imaging technology, hyperspectral data in different spectral ranges, such as visible and near-infrared, short-wave infrared, etc., can be acquired simultaneously. Data fusion between these nonoverlapping spectral ranges’ hyperspectral data has become an urgent task. Most existing hyperspectral data fusion methods focus on two types of hyperspectral data with overlapping spectral ranges, requiring spectral response functions as a necessary condition, which is not applicable to this task. To address this issue, we propose the combined spectral unmixing fusion (CSUF) method, an unsupervised method with certain physical significance. It effectively solves the problem of hyperspectral data fusion with nonoverlapping spectral ranges through the two hyperspectral data point spread function estimation and combined spectral unmixing. Experiments on airborne datasets and HJ-2 satellite data show that, compared with various leading methods, our method achieves the best performance in terms of reference evaluation indicators such as the PSNR and SAM, as well as the non-reference evaluation indicator the QNR. Furthermore, we deeply analyze the spectral response relationship and the impact of the ratio of spectral bands between the fused data on the fusion effect, providing references for future research.

Compared with traditional single-view and radiometric-only observations, multiangle polarimetric observations of water-leaving radiation play a crucial role in enhancing the retrieval of ocean constituents and aerosol microphysical properties. In this study, the impacts of uncertainties in the degree of polarization (DOP) of water-leaving radiance (L_w) on the retrieval of oceanic constituents and inherent optical properties (IOPs) were investigated via global radiative transfer (RT) simulations and the fully connected U-Net (FCUN) model. The uncertainties in the retrieval of oceanic constituents and IOPs were further investigated with various sensor azimuth angles. The results indicated that the global mean absolute percentage errors (MAPEs) for differing oceanic constituents and IOPs significantly decreased as the number of observation angles increased. Taking the retrieval of Chla as an example, the global MAPEs between the FCUN predictions and RT simulation inputs for Chla concentrations under differing observation angles were 7.41%, 3.76%, 2.70%, 2.44%, 2.62%, and 1.82%. Moreover, the MAPEs at sensor azimuth angles of 0° and 30° were significantly lower than those at other azimuth angles for the single-view observations. As the number of observation angles increased, the variation in MAPEs with the sensor azimuth angle gradually weakened. Furthermore, the impact of errors in the L_w DOP on the retrieval uncertainties decreased as the number of observation angles increased, and the global MAPEs of Chla after adding the various random instrument noises were 46.56% (46.91%), 6.59% (7.21%), 5.21% (5.79%), 4.72% (4.98%), 3.99% (4.52%), and 3.64% (4.03%). Overall, the multiangle polarimetric observations can suppress or balance the impact of uncertainties in the L_w DOP on the retrieval of oceanic constituents and IOPs.