In recent years, we have witnessed significant development in the space sector, in particular regarding Earth imaging. Small satellites, whose size and construction make their production much cheaper, are becoming increasingly popular. As a result, a larger number of satellites may be placed in space, and thus, they may perform more frequent observations of selected spots on Earth. Unfortunately, the construction of these satellites also affects their observation capacity as they have a weaker spatial resolution. Scientists have been dealing with the problem of improving the spatial resolution of satellite imaging for many years. Numerous methods were developed that allow for the best possible representation of high-resolution images based on low-resolution images. However, the application of traditional solutions to improve the resolution of digital images requires an additional high-resolution image. As far as images obtained by small satellites (e.g., nano, micro, or mini) are concerned, the difference between the spatial resolution of panchromatic and multispectral images is small (e.g., for SkySat-3 – SkySat-15 satellites, it is only 0.16 m). The need to increase the spatial resolution of an image that does not have a corresponding higher resolution image (e.g., a panchromatic image or a sequence of images) causes additional problems. This article presents a review of the methods to improve the spatial resolution of small-satellite imaging. The authors analyze the interpolation, pansharpening, and digital image processing methods. Additionally, the article focuses on presenting solutions based on deep learning that enables the enhancement of the spatial resolution of images obtained from small satellites. The methodology of creating databases used for network training is described. Finally, the authors present the main limitations of the analyzed solutions and future development trends that will enable to improve the spatial resolution with the use of a single image.

With the dynamic technological development, we are witnessing a major progress in solutions that allow for the observation of Earth's surface. The constructed satellites are becoming still smaller and lighter. Small satellites (those weighing less than 100kg) have a significant drawback (in terms of observational capabilities). Due to their limitations, the installed optic systems are not perfect. As a result, the quality of the obtained images is lower, including lower resolution, although the satellites move on the Low Earth Orbit. In the case of images lacking a high-resolution counterpart, the spatial resolution of the imagery can be improved using single-image super-resolution (SISR) algorithms. In this article, we present an SISR solution based on a new network called MCWESRGAN, which is a modification of the popular ESRGAN network. We propose a novel strategy that introduces a multi-column discriminator model. The generator model is trained using Wasserstein loss. The introduced modifications enable a tenfold reduction in the training time of the network. The proposed algorithm is verified using images obtained from space, aerial imagery, and the Dataset for Object deTection in Aerial Images (DOTA) database. A set of evaluation methods for super-resolution (SR) images is proposed to verify the results. These evaluation methods indicate areas that are poorly estimated by the algorithm. Furthermore, as part of the conducted experiments, an absolute assessment method for interpretational potential based on the power spectral density of the image (PSD) is proposed, allowing for determining the magnitude of interpretational improvement after applying resolution enhancement algorithms. The conducted research demonstrates that we achieve better qualitative and quantitative results than classical ESRGAN methods and other state-of-the-art (SOTA) approaches.

Dynamic technological progress has contributed to the development of systems imaging of the Earth’s surface as well as data mining methods. One such example is super-resolution (SR) techniques that allow for the improvement of the spatial resolution of satellite imagery on the basis of a low-resolution image (LR) and an algorithm using deep neural networks. The limitation of these solutions is the input size parameter, which defines the image size that is adopted by a given neural network. Unfortunately, the value of this parameter is often much smaller than the size of the images obtained by Earth Observation satellites. In this article, we presented a new methodology for improving the resolution of an entire satellite image, using a window function. In addition, we conducted research to improve the resolution of satellite images acquired with the World View 2 satellite using the ESRGAN network, we determined the number of buffer pixels that will make it possible to obtain the best image quality. The best reconstruction of the entire satellite imagery using generative neural networks was obtained using a Triangular window (for 10% coverage). The Hann-Poisson window worked best when more overlap between images was used.

Processing single high-resolution satellite images may provide a lot of important information about the urban landscape or other applications related to the inventory of high-altitude objects. Unfortunately, the direct extraction of specific features from single satellite scenes can be difficult. However, the appropriate use of advanced processing methods based on deep learning algorithms allows us to obtain valuable information from these images. The height of buildings, for example, may be determined based on the extraction of shadows from an image and taking into account other metadata, e.g., the sun elevation angle and satellite azimuth angle. Classic methods of processing satellite imagery based on thresholding or simple segmentation are not sufficient because, in most cases, satellite scenes are not spectrally heterogenous. Therefore, the use of classical shadow detection methods is difficult. The authors of this article explore the possibility of using high-resolution optical satellite data to develop a universal algorithm for a fully automated estimation of object heights within the land cover by calculating the length of the shadow of each founded object. Finally, a set of algorithms allowing for a fully automatic detection of objects and shadows from satellite and aerial imagery and an iterative analysis of the relationships between them to calculate the heights of typical objects (such as buildings) and atypical objects (such as wind turbines) is proposed. The city of Warsaw (Poland) was used as the test area. LiDAR data were adopted as the reference measurement. As a result of final analyses based on measurements from several hundred thousand objects, the global accuracy obtained was ±4.66 m.

Inpainting is a technique that allows for the reconstruction of images and the removal of unnecessary elements. In our research, we employed inpainting to eliminate erroneous lines in the images and examined its abilities in improving classification quality. To reduce the erroneous lines, we designed ResGMCNN, whose multi-column generator model uses residual blocks. For our studies, we used data from the COWC and DOTA datasets. The GMCNN model with residual connections outperformed most classical inpainting methods, including the Telea and Navier–Stokes methods, achieving a maximum structural similarity index measure (SSIM) of 0.93. However, despite the improvement in filling quality, these results still lag behind the Criminisi method, which achieved the highest SSIM values (up to 0.99). We investigated the improvement in classification quality by removing vehicles from the road class in images acquired by UAVs. For vehicle removal, we used Criminisi inpainting, as well as Navier–Stokes and Telea for comparison. Classification was performed using eight classifiers, six of which were based on machine learning, where we proposed our solutions. The results showed that classification quality could be improved by several to over a dozen percent, depending on the metric, image, and classification method. The F1-score and Cohen Kappa metrics indicated an improvement in classification quality of up to 13% in comparison to the classification of the original image. Nevertheless, each of the classical inpainting methods examined improved the road classification.

Image inpainting refers to the process of restoring missing or damaged areas in an image. This research field has been very active in recent years, driven by various applications such as reconstructing lost fragments, concealing data loss in corrupted image transmissions, removing objects in image editing, and interpolating image content for reconstruction in image-based rendering (IBR) from various fields of view. This paper presents existing methods of image inpainting, covering classical approaches, CNN-based methods, and GAN-based methods. Additionally, it explores techniques related to steganography, adversarial image synthesis, and false image generation. Examples of applications are provided for each category of image modification methods. Although image inpainting and digital camouflage are not yet widely studied in the remote sensing community, there has been a growing interest in these topics in recent years. To broaden the understanding of these methods, this study also reviews techniques developed in the field of computer science, which have the potential to be adapted for remote sensing applications. The main contribution of this paper is the presentation of various forms of digital masking, extending beyond traditional inpainting. We also provide a curated list of publicly available datasets that can support the development of new solutions, along with a selection of qualitative metrics for the robust evaluation of image inpainting algorithms. The code are available at https://github.com/KK-MUT/DigitalCamouflage