This paper proposes a systematic image mosaicking methodology to produce hyperspectral image for environment monitoring using an emerging UAV-based push-broom hyperspectral imager. The suitability of alternative methods in each step is assessed by experiments of an urban scape, a river course and a forest study area. First, the hyperspectral image strips were acquired by sequentially stitching the UAV images acquired by push-broom scanning along each flight line. Next, direct geo-referencing was applied to each image strip to get initial geo-rectified result. Then, with ground control points, the curved surface spline function was used to transform the initial geo-rectified image strips to improve their geometrical accuracy. To further remove the displacement between pairs of image strips, an improved phase correlation (IPC) and a SIFT and RANSAC-based method (SR) were used in image registration. Finally, the weighted average and the best stitching image fusion method were used to remove the spectral differences between image strips and get the seamless mosaic. Experiment results showed that as the GCPs‘ number increases, the mosaicked image‘s geometrical accuracy increases. In image registration, there exists obvious edge information that can be accurately extracted from the urban scape and river course area; comparative results can be achieved by the IPC method with less time cost. However, for the ground objects with complex texture like forest, the edges extracted from the image is prone to be inaccurate and result in the failure of the IPC method, and only the SR method can get a good result. In image fusion, the best stitching fusion method can get seamless results for all three study areas. Whereas, the weighted average fusion method was only useful in eliminating the stitching line for the river course and forest areas but failed for the urban scape area due to the spectral heterogeneity of different ground objects. For different environment monitoring applications, the proposed methodology provides a practical solution to seamlessly mosaic UAV-based push-broom hyperspectral images with high geometrical accuracy and spectral fidelity.

This paper designs an object-based confusion matrix (OCM) classification accuracy assessment scheme to accurately estimate the overall and individual category classification accuracy. The estimation protocol and the sample data collection procedures are both taken into account. On the one hand, the two commonly used OCM construction methods based on object element and weighted by object area are analyzed, which indicate the classifier's distinguish ability and the thematic map accuracy respectively. With consideration that the object location uncertainty introduces the reference category uncertainty and may lead to the bias of thematic map accuracy assessment result, a novel fuzzy OCM construction method is proposed to more accurately assess the classification accuracy. On the other hand, a simple sample data collection strategy is proposed and validated to collect representative accuracy assessment samples. The object-oriented Quickbird image land use classification accuracy assessment experiment results are analyzed to validate the applicability of the proposed schemes. Suggestions on how to use object-based confusion matrix method in classification accuracy assessment are given in conclusion.

A water quality parameter retrieval scheme based on the UAV push-broom hyperspectral images was designed and validated for assessing the ecological health of Zhang Wei Xin River in Dezhou distinct, China. First, a UAV carrying a push-broom hyperspectral imager that is lightweight and has a small size was used to acquire high spatial and hyperspectral resolution images. Then, the mosaicked reflectance data of the whole river were produced by a seamless image mosaicking method with high geometrical accuracy and spectral fidelity. Next, the in-field measurements of different parameters and the corresponding spectral reflectance from the mosaicked images at the sampling points were used to build the water quality parameter retrieval models for total phosphorus (TP), chlorophyll a (Chla), and total suspended solids (TSS). To validate the model, the retrieval results of the testing sampling points were compared with the measured parameters. The coefficients of determination R² of TP, Chla, and TSS were 0.886, 0.918, and 0.968, respectively. The retrieved TP, Chla, and TSS maps showed that the water pollution of Zhang Wei Xin River is serious, the total phosphorus exceeds the standard, and the water body is in a state of eutrophication. The UAV-based hyperspectral remote sensing technique provides a cost-effective method for inland water monitoring at a local scale with high accuracy.

Hyperspectral (HS) imaging has achieved breakthroughs in many applications, such as remote sensing and object recognition. However, the spatial resolution of HS images is still insufficient due to the limitations of sensor technology and cost. In this article, we propose an HS image superresolution method that combines low-resolution (LR) HS images and high-resolution (HR) panchromatic (PAN) images. To exploit the spectral signatures in the LR-HS images while introducing details from the HR-PAN images during the image fusion procedure, an image matting model is used to fuse the original LR-HS images and the HR-PAN images. Specifically, to preserve the spectral components during the fusion procedure, two different alpha channels in the image matting model are generated based on the HS and PAN image structure tensors, which suppress spectral distortion and improve the quality of the reconstructed HR-HS image. Experimental results based on public datasets demonstrate the advantage of our proposed method in both preserving spectral information and enhancing HS image spatial resolution.

Under night vision conditions, visible-spectrum images often fail to capture background details. Conventional visible and infrared fusion methods generally overlay thermal signatures without preserving latent features in low-visibility regions. This paper proposes a novel deep learning-based fusion algorithm to enhance visual perception in night driving scenarios. Firstly, a local adaptive enhancement algorithm corrects underexposed and overexposed regions in visible images, thereby preventing oversaturation during brightness adjustment. Secondly, ResNet152 extracts hierarchical feature maps from enhanced visible and infrared inputs. Max pooling and average pooling operations preserve critical features and distinct information across these feature maps. Finally, Linear Discriminant Analysis (LDA) reduces dimensionality and decorrelates features. We reconstruct the fused image by the weighted integration of the source images. The experimental results on benchmark datasets show that our approach outperforms state-of-the-art methods in both objective metrics and subjective visual assessments.

Hyperspectral imaging systems are widely used in precision agriculture, environmental monitoring, and mineral exploration. However, current systems often suffer from high cost, large size and weight, and considerable system complexity, which hinder their widespread deployment. To overcome these limitations and achieve a better balance between performance, cost, and portability, this work aims to develop a compact, cost-effective visible-to-near-infrared (VNIR, 400–1000 nm) hyperspectral camera based on Schwarzschild configuration and commercial off-the-shelf (COTS) components. The development followed a comprehensive methodology encompassing theoretical design, simulation, prototype assembly, and performance testing. The all-reflective optical system effectively eliminates chromatic aberration and minimizes energy loss, achieving an integration time as short as several milliseconds and a push-broom frame rate of 200 fps. The optical design leveraged optical path length theory and the unobscured Schwarzschild structure to optimize off-axis mirrors and a plane grating. Optical performance was optimized and verified using simulations, which confirmed that spot sizes at all field positions were highly concentrated and that critical distortions such as smile and keystone were controlled within several pixels. A prototype was assembled on a precision optical bench using multi-axis adjustable mounts and then integrated into a precisely machined housing, achieving a total weight less than 2 kg. Calibration verified a spectral coverage of 400–1000 nm and a resolution of 5 nm. Imaging experiments demonstrated the system’s ability to resolve subtle spectral features, successfully distinguishing different vegetations and artificial materials based on their spectral signatures—particularly the strong NIR (780–1000 nm) reflectance of vegetation versus synthetic green materials. The camera offers a high-performance, low-cost solution suitable for applications including precision agriculture, environmental monitoring, mineral exploration, and others.