Snow-free Image Research Articles

Multi-Temporal Snow-Covered Remote Sensing Image Matching via Image Transformation and Multi-Level Feature Extraction

To address the challenge of image matching posed by significant modal differences in remote sensing images influenced by snow cover, this paper proposes an innovative image transformation-based matching method. Initially, the Pix2Pix-GAN conversion network is employed to transform remote sensing images with snow cover into images without snow cover, reducing the feature disparity between the images. This conversion facilitates the extraction of more discernible features for matching by transforming the problem from snow-covered to snow-free images. Subsequently, a multi-level feature extraction network is utilized to extract multi-level feature descriptors from the transformed images. Keypoints are derived from these descriptors, enabling effective feature matching. Finally, the matching results are mapped back onto the original snow-covered remote sensing images. The proposed method was compared to well-established techniques such as SIFT, RIFT2, R2D2, and ReDFeat and demonstrated outstanding performance. In terms of NCM, MP, Rep, Recall, and F1-measure, our method outperformed the state of the art by 177, 0.29, 0.22, 0.21, and 0.25, respectively. In addition, the algorithm shows robustness over a range of image rotation angles from −40° to 40°. This innovative approach offers a new perspective on the task of matching multi-temporal snow-covered remote sensing images.

Snow Mask Guided Adaptive Residual Network for Image Snow Removal

Image restoration under severe weather is a challenging task. Most of the past works focused on removing rain and haze phenomena in images. However, snow is also an extremely common atmospheric phenomenon that will seriously affect the performance of high-level computer vision tasks, such as object detection and semantic segmentation. Recently, some methods have been proposed for snow removing, and most methods deal with snow images directly as the optimization object. However, the distribution of snow location and shape is complex. Therefore, failure to detect snowflakes/snow streak effectively will affect snow removing and limit the model performance. To solve these issues, we propose a Snow Mask Guided Adaptive Residual Network (SMGARN). Specifically, SMGARN consists of three parts, Mask-Net, Guidance-Fusion Network (GF-Net), and Reconstruct-Net. Firstly, we build a Mask-Net with Self-pixel Attention (SA) and Cross-pixel Attention (CA) to capture the features of snowflakes and accurately localized the location of the snow, thus predicting an accurate snow mask. Secondly, the predicted snow mask is sent into the specially designed GF-Net to adaptively guide the model to remove snow. Finally, an efficient Reconstruct-Net is used to remove the veiling effect and correct the image to reconstruct the final snow-free image. Furthermore, we propose a more refined dataset of real snow images, SnowWorld24, to provide faster evaluation of snow-free images. Extensive experiments show that our SMGARN numerically outperforms all existing snow removal methods, and the reconstructed images are clearer in visual contrast. All codes are available at https://github.com/MIVRC/SMGARN.

Snowed autoencoders are efficient snow removers

Snow is a harsh natural phenomenon that greatly affects the performance of advanced computer vision tasks. Recently, most image desnowing methods rely on complex model structures, leading to increased carbon emissions and impossibilities to deploy on lightweight computing devices. Additionally, most deep learning-based methods do not effectively utilize the position information of snow particles, which will limit the performance of the model. To address these issues, considering that snow particles have block-shaped shape distribution and color uniformity similar to a mask, we propose a snowed autoencoder (SAE) desnowing method. Specifically, the SAE is composed of four parts: snowed masking process, SAE encoder, SAE decoder, and prediction stage. First, a snowy image is passed through the snowed masking process to generate a mask that exactly covers the snow particles and output the coordinates of the four vertices of the mask and the index of each segmented image patch. The input image and each mask coordinate and index are passed through the SAE encoder based on the snow particle attention module to generate an image representation for identifying typical features of snow particles. Then, the generated typical features of snow particles are fed into the SAE decoder to reconstruct image patches without snow. Finally, all the patches will be passed through the prediction stage to reconstruct the original size of the snow-free image. A large number of experiments show that the proposed SAE desnowing method achieves the state-of-the-art desnowing performance on three synthetic and one real-world desnowing datasets. The images after desnowing by SAE have better color details and are more consistent with human visual habits. In addition, SAE has faster desnowing speed and fewer parameters. The code is available at https://github.com/awhitewhale/SAE.

Ground Displacements in NY Using Persistent Scatterer Interferometric Synthetic Aperture Radar and Comparison of X- and C-Band Data

In this study, we investigated the quality of Interferometric Synthetic Aperture Radar (InSAR) data to measure surface displacements in upstate New York, an area with dense vegetation, snowy winters, and strong seasonal signals. We used data from the German Space Agency’s TerraSAR-X and TanDEM-X satellites (X-band, 3.1 cm radar wavelength) as well as the European Space Agency’s Sentinel-1 satellite (C-band, 5.6 cm radar wavelength); both datasets covered a ~3-year time period from 2018 to 2021. Using persistent scatterer interferometry (PSI), we were able to observe several deforming features in the region with sub-centimeter/year deformation rates. We also examined a version of the X-band data that we spatially averaged to the same pixel size as the Sentinel-1 imagery in order to separate out the effects of wavelength and pixel size on PSI accuracy and coverage. Overall, the largest number of stable PS points was found in the full-resolution X-band data, which was followed by the C-band data and then by the downsampled X-band data. Our analysis also included a subset of snow-free imagery so that we could assess the effect that snow-covered images had on the distribution and accuracy of PS points and the resulting time series. This analysis revealed that PS populations increased by 50–60% for the snow-free data when compared with analyses using the full datasets. The average deformation rates inferred from the time series generated using only snow-free images were nearly identical to those estimated from the full time series. We assessed the accuracy of the inferred rates through comparisons between the results of different datasets and with limited ground survey data. We found that all of the inferred deformation rates from each of the datasets agreed with in situ measurements in an area of known ground subsidence above an underground salt mine in Lansing, NY. The S1 datasets, however, had higher levels of noise.

Removing Snow From a Single Image Using a Residual Frequency Module and Perceptual RaLSGAN

Removing snow particles from an image is a complicated task due to the particles’ shape, size, and color. The latest snow removal methods remove snow from a single image but retain some snow and salt-and-pepper particles. Some approaches, while trying to remove snow from a single image, produce blurry artifacts. In this paper, we solve these problems by designing a network model that consists of a residual generative network, a snow-free image generative network, and a perceptual relativistic discriminative network. In both generative networks, we assign the residual frequency network (ReFNet) as our bottleneck module. Our network model learns to map two relationships. First, the input snowy image is trained to map the snow mask image in the dataset. Then, a retained image resulting from subtraction between an input image and the estimated residual image is concatenated with the input snowy image and mapped to the desired snow-free ground truth. Moreover, we use a perceptual identical-paired adversarial network based on a relativistic discriminative network to make our training results more robust. Our results achieve greater performance than state-of-the-art methods on both synthetic and real-world snowy images.

Recession of Gya Glacier and the 2014 glacial lake outburst flood in the Trans-Himalayan region of Ladakh, India

This study assessed spatiotemporal changes at Gya Glacier, the associated development of a proglacial lake, and reconstructed the 2014 outburst flood that struck Gya Village in the Trans-Himalayan region of Ladakh, India. This study analyzed and for the first time modeled a Glacial Lake Outburst Flood (GLOF) event in the Trans-Himalayan region of Ladakh. Glacier and glacial lakes changes were quantified using remote sensing data supplemented with field observations. Glacier ice-thickness and glacier-bed overdeepenings were modeled using a shear-stress based model, GlabTop (Glacier-bed Topography). The reconstruction of the 2014 GLOF and the potential hazard assessment of Gya Lake were carried out using the hydrodynamic model HEC-RAS; results were validated against ground-collected data. Temporal evaluation of satellite data revealed a 45.6% loss in the total glacier area between 1969 and 2019. The earliest snow-free image available for the region shows that a proglacial lake existed as early as 1969 with an area of 3.06 ha. The lake has expanded to ~11 ha in 2019. Results from the GlabTop model suggest that the lake could grow further up to 12 ha in the future. Field-based geomorphic indicators suggest that the 2014 GLOF event resulted from a piping failure of the frontal moraine destroying numerous agricultural fields, some buildings, downstream infrastructure, and eroded natural channel embankments. The reconstruction of the event revealed that 25% of the lake waters drained out with a peak discharge of 470 m3s−1, inundating an area of ~4 km2 around Gya Village. However, a complete breaching of the terminal moraine could result in an event that would be 5.5 times larger than the 2014 GLOF. Therefore, this study could be useful not only in planning disaster-resilient infrastructure around proglacial lake environments in the cold-arid Ladakh but also in framing mitigation plans to reduce risk for vulnerable downstream communities.

Single Image Snow Removal Using Sparse Representation and Particle Swarm Optimizer

Images are often corrupted by natural obscuration (e.g., snow, rain, and haze) during acquisition in bad weather conditions. The removal of snowflakes from only a single image is a challenging task due to situational variety and has been investigated only rarely. In this article, we propose a novel snow removal framework for a single image, which can be separated into a sparse image approximation module and an adaptive tolerance optimization module. The first proposed module takes the advantage of sparsity-based regularization to reconstruct a potential snow-free image. An auto-tuning mechanism for this framework is then proposed to seek a better reconstruction of a snow-free image via the time-varying inertia weight particle swarm optimizers in the second proposed module. Through collaboration of these two modules iteratively, the number of snowflakes in the reconstructed image is reduced as generations progress. By the experimental results, the proposed method achieves a better efficacy of snow removal than do other state-of-the-art techniques via both objective and subjective evaluations. As a result, the proposed method is able to remove snowflakes successfully from only a single image while preserving most original object structure information.

Stacked dense networks for single-image snow removal

Single image snow removal is important since snowy images usually degrade the performance of computer vision systems. In this paper, we deduce a physics-based snow model and propose a novel snow removal method based on the snow model and deep neural networks. Our model decomposes a snowy image into a nonlinear combination of a snow-free image and dynamic snowflakes. Inspired by our model and DenseNet connectivity pattern, we design a novel Multi-scale Stacked Densely Connected Convolutional Network (MS-SDN) to simultaneously detect and remove snowflakes in an image. The MS-SDN is composed of a multi-scale convolutional sub-net for extracting feature maps and two stacked modified DenseNets for snowflakes detection and removal. The snowflake detection sub-net guides snow removal through forward transmission, and the snowflake removal sub-net adjusts snow detection through back transmission. In this way, snowflake detection and removal mutually improve the final results. For training and testing our method, we constructed a large-scale benchmark synthesis dataset which contains 3000 triplets of snowy images, snowflakes, and snow-free images. Specifically, the snow-free images are captured from snow scenes, and the snowy images are synthesized by using our deduced snow model. Our extensive quantitative and qualitative experimental results show that our MS-SDN performs better than several state-of-the-art methods, and the stacked structure is better than multi-branch structures in terms of snow removal.

Boreal canopy surfaces from spaceborne stereogrammetry

Surface elevation estimates from high resolution spaceborne image (HRSI) stereogrammetry are used to examine fine-scaled structure of boreal forest canopies. These data can depict detailed spatial patterns of vertical forest structure at remote sites across the circumpolar domain where these estimates would otherwise be unavailable. This work examines where these estimates are most effective at describing vertical forest structure to explain which canopy surfaces they represent. We evaluated the variation in canopy surface estimates captured from four general types of HRSI digital surface models (DSMs) across the full range of boreal canopy cover. These DSMs, classified into 4 types by grouping them according to the acquisition's (1) sun elevation angle (low or high) and (2) seasonality-driven ground surface condition (snow presence/absence), vary with acquisition characteristics and the details of this variation continues to be studied. We explored some of this variation by comparing the distributions of differences in boreal canopy percentile heights derived from reference small footprint lidar in Tanana Valley, Alaska with canopy surface elevations derived from these 4 types of HRSI DSMs. We examined how canopy surface estimates from HRSI DSMs differ according to acquisition characteristics and canopy cover, and ultimately which canopy surfaces are represented in these DSMs. Our results help clarify which boreal canopy surfaces are representative of those captured with HRSI DSMs. They show that in the Tanana Valley (1) DSMs grouped by sun elevation angle and ground surface condition provide different surface estimates of boreal canopies; (2) the two DSM types that appear to most differently capture boreal forest canopy surfaces are DSMs from snow-free images acquired at sun elevation angles <30° (Low sun elev. & snow-free) and those with snow-cover at sun elevation angles ≥30° (High sun elev. & snow-free); (3) DSMs with snow most often do not capture upper canopy surfaces; (4) the “Low sun elev. & snow-free” DSMs resolve surfaces that are most representative of upper canopy surfaces (dense forests >60% cover, 70th–80th percentile heights); and (5) in the most dense forests (>80% cover) where canopy gaps are least likely to bias downward the average surface estimates, the snow-free DSM types are representative of 70th - 80th percentile heights (“Low sun elev. & snow-free”) and 60th–70th percentile heights (“High sun elev. & snow-free”). The combination of horizontal structure (canopy cover) and acquisition characteristics affect the boreal vertical structure (canopy surface height) estimates from spaceborne stereogrammetry. These effects should be considered when analyzing products derived from HRSI DSMs, and as part of a comprehensive approach to spaceborne remote sensing of circumpolar boreal forests.

DesnowNet: Context-Aware Deep Network for Snow Removal.

Existing learning-based atmospheric particle-removal approaches such as those used for rainy and hazy images are designed with strong assumptions regarding spatial frequency, trajectory, and translucency. However, the removal of snow particles is more complicated because they possess additional attributes of particle size and shape, and these attributes may vary within a single image. Currently, hand-crafted features are still the mainstream for snow removal, making significant generalization difficult to achieve. In response, we have designed a multistage network named DesnowNet to in turn deal with the removal of translucent and opaque snow particles. We also differentiate snow attributes of translucency and chromatic aberration for accurate estimation. Moreover, our approach individually estimates residual complements of the snow-free images to recover details obscured by opaque snow. Additionally, a multi-scale design is utilized throughout the entire network to model the diversity of snow. As demonstrated in the qualitative and quantitative experiments, our approach outperforms state-of-the-art learning-based atmospheric phenomena removal methods and one semantic segmentation baseline on the proposed Snow100K dataset. The results indicate our network would benefit applications involving computer vision and graphics.

Single image rain and snow removal via guided L0 smoothing filter

Since no temporal information can be exploited, rain and snow removal from single image is a challenging problem. In this paper, an improved rain and snow removal method from single image is proposed by designing a guided L0 smoothing filter. The designed filter is inspired by the previous L0 gradient minimization. Then a coarse rain-free or snow-free image can be obtained with the proposed filter, and the final refined result is recovered by a further minimization operation depending on the observed image. Experimental results show that the proposed algorithm generates better or comparable outputs than the state-of-the-art algorithms in rain and snow removal task for single image.

Interpretation of a RADARSAT-2 fully polarimetric time-series for snow cover studies in an Alpine context – first results

The aim of the SOAR #1341 project is to perform temporal analyses of changes in RADARSAT-2 full-polarimetry parameters on snow cover in a mountainous area. The objective of the present study was to determine whether there is a correlation between changes in radar statistics and changes in physical snow parameters during winter and spring. This paper focuses on the preprocessing of the images and presents the methodological steps and first results obtained in full polarimetry mode. Six RADARSAT-2 quad-pol images were acquired between January 2009 and January 2010, five in different snow conditions and one snow-free image in summer used as a reference. The fine acquisition mode was selected with a medium incidence angle (39°). A combination of LANDSAT-7 and SPOT optical images and field measurements was used for the validation step. First, RADARSAT-2 images had to be pre-processed due to the influence of high mountain topography on the polarimetric signal: the planned incidence angle was computed using a fine digital elevation model (DEM). Next, the DEM and optical dataset were reprojected onto the slant range mode of RADARSAT-2, the configuration required to preserve the phase signal and polarimetric statistics. Polarimetric analysis was performed using the PolSARpro software from ESA/IETR. The coherency matrices were calculated for each RADARSAT-2 image. Polarimetric descriptors based on the eigenvector–eigenvalue decomposition theorem of this coherency matrix were obtained. The behavior of the Single Eigenvalue Relative Difference, a polarimetric parameter that depends on scattering mechanisms, was analyzed. The polarimetric analysis showed an increase in the multiple scattering mechanism with a dry snow cover compared with the snow-free image. With a wet snow cover, there was an increase in contribution of the single scattering compared with the snow-free image. Our results showed that it was possible to identify temporal changes in dry, wet, or no snow characteristics throughout the winter season by analyzing primary polarimetric decomposition parameters and by comparing them with measurements made at 10 field sites.