VHR Satellite Imagery Research Articles

Seals from outer space - Population census of southern elephant seals using VHR satellite imagery

This work aims to assess the utility of World-View 3 satellite imagery of 31 cm resolution for conducting population censuses of southern elephant seals (SES). For the first time, we present a comparison of the results of image analyses of a 31 cm resolution satellite image and 1.39 cm resolution drone-based orthophoto of a SES breeding colony located at Patelnia Point (King George Island, breeding season 2019). The time distance separating both visualisations of the same area was 15 h 4 min, which allowed us to determine the female recognition from the VHR image with an error of 1.5–7.0%, depending on the density of females in the breeding formation. An important component of this error is false negative matches, which can have a significant impact when selecting the objects needed to train deep learning models. In addition, the pup recognition error was much higher and was 24.6–40.2%, which depended on the density of animals in harems, a large variation in the size of the pups and the diverse substrate on which they were deposited. Counting on VHR images was performed manually by eleven observers familiar with the morphological characteristics of SES. The resolution of the VHR photo allows the identification of females and males involved in breeding due to the clear dimorphism of their sizes. We recommend using VHR satellite imagery to determine the number of breeding formations and spatial relationships between them, as well as for counting females and males during the breeding haul-out. The resolution of satellite imagery available at this time is still insufficient to be used to determine breeding success. With increasing access to satellite resources and an ever-expanding database of VHR images, this is a promising tool to fill the gaps in knowledge about the global population of SESs.

Digital surface model generation from high-resolution satellite stereo imagery based on structural similarity

In this study, we proposed an automatic approach to extract DSMs from VHR satellite imagery. This study aims to apply the histogram of orientated phase congruency (HOPC) descriptor and SIFT-Flow algorithm to match multi-view VHR satellite images densely and extract detailed DSMs competitive with ones generated from Light Detection and Ranging (LiDAR) point clouds. The HOPC descriptor relies on detecting features by calculating the phase congruency model, which allows the image features to be highly robust against illumination changes. Moreover, we developed an automatic algorithm to extract precise and homogenously distributed tie-points and subsequently conduct bias correction of the Rational Polynomial Coefficients (RPC). We tested our proposed algorithm on four WorldViw-3 datasets captured from different viewpoints, and compared the results with the DSMs obtained by the Semi-Global Matching (SGM) algorithm. The experimental results indicated the proficiency of our proposed method in generating accurate and reliable DSMs using VHR satellite data.

Automatic slums identification around normal and smart cities: using Machine-learning on VHR Satellite Imagery

Automatic slums identification around normal and smart cities: using Machine-learning on VHR Satellite Imagery

The Integration of VHR Satellite Imagery, GPR Survey and Boring for Archaeological Prospection at the Longcheng Site in Anhui Province, China

This paper is focused on the joint use of non‐invasive and minimal intervention techniques for supporting archaeological prospection. Very high resolution (VHR) satellite imagery analysis and ground‐penetrating radar (GPR) and boring surveys were integrated for the study of the Longcheng site, located near Hefei city in Anhui Province, China, to test their effectiveness and efficiency in prospecting archaeological remains and evaluating their degree of preservation. First, target locations of potential archaeological structures were identified on a WorldView‐2 (WV‐2) satellite image through spatial and radiometric enhancement, interpretation and object‐oriented classification. Second, archaeological features extracted from the WV‐2 imagery were further investigated by a GPR survey that provided detailed cross‐checking information about buried remains. Finally, a subsequent boring survey was conducted across those prospective archaeological structures in order to map the stratigraphic sequences on the basis of colour, compactness and the inclusions contained in the soil, and then to test their correspondence with the GPR data. The boring led to detailed confirmation of the results produced by the remote sensing analyses and GPR surveys, as well as the discovery of datable artefacts. On the basis of all the integrated data, the preliminary layout and structure of the Longcheng site was reconstructed in GIS. Furthermore, the widths, lengths, heights and burial depths of these buried archaeological structures were estimated in detail.

Automatic Extraction of Arbitrarily Shaped Buildings from VHR Satellite Imagery

Automatic Extraction of Arbitrarily Shaped Buildings from VHR Satellite Imagery

Automated retrieval of forest structure variables based on multi-scale texture analysis of VHR satellite imagery

The main goal of this study is to design a method to describe the structure of forest stands from Very High Resolution satellite imagery, relying on some typical variables such as crown diameter, tree height, trunk diameter, tree density and tree spacing. The emphasis is placed on the automatization of the process of identification of the most relevant image features for the forest structure retrieval task, exploiting both spectral and spatial information. Our approach is based on linear regressions between the forest structure variables to be estimated and various spectral and Haralick’s texture features. The main drawback of this well-known texture representation is the underlying parameters which are extremely difficult to set due to the spatial complexity of the forest structure.To tackle this major issue, an automated feature selection process is proposed which is based on statistical modeling, exploring a wide range of parameter values. It provides texture measures of diverse spatial parameters hence implicitly inducing a multi-scale texture analysis. A new feature selection technique, we called Random PRiF, is proposed. It relies on random sampling in feature space, carefully addresses the multicollinearity issue in multiple-linear regression while ensuring accurate prediction of forest variables. Our automated forest variable estimation scheme was tested on Quickbird and Pléiades panchromatic and multispectral images, acquired at different periods on the maritime pine stands of two sites in South-Western France. It outperforms two well-established variable subset selection techniques. It has been successfully applied to identify the best texture features in modeling the five considered forest structure variables. The RMSE of all predicted forest variables is improved by combining multispectral and panchromatic texture features, with various parameterizations, highlighting the potential of a multi-resolution approach for retrieving forest structure variables from VHR satellite images. Thus an average prediction error of ∼1.1m is expected on crown diameter, ∼0.9m on tree spacing, ∼3m on height and ∼0.06m on diameter at breast height.

Wildfire hazard in the home ignition zone: An object-oriented analysis integrating LiDAR and VHR satellite imagery

Many spatially explicit studies of wildfire hazard focus on the wildland–urban interface (WUI), the area where natural vegetation intersects or mixes with structures. However, research suggests that the characteristics of a small portion of the WUI, the home ignition zone, largely determine potential for ignition from wildfire. The home ignition zone (HIZ) is the area that includes a structure and its surroundings out to 30–60 m. The primary goal of this study is to develop metrics to characterize land cover, burned area, and topography in the HIZ. Pre-fire metrics (i.e. related to land cover and topography) help identify relatively hazardous individual HIZes or neighborhoods of HIZes. Post-fire metrics can be used to assess the burned area across land cover types, inside and outside the HIZ. To calculate the HIZ metrics, multiple data sources (e.g. high resolution 8-band multispectral imagery and LiDAR point clouds) were integrated using an object-oriented image analysis. The setting for the study is the Fourmile Canyon area west of Boulder, Colorado, a data-rich area which experienced a large, destructive wildfire in September 2010. The land cover, burn area, and topography metrics were successfully and accurately calculated and then pre-fire metrics were combined into a simple HIZ hazard index. HIZ characteristics broadly mirror the characteristics of the WUI within the fire perimeter as a whole, though the HIZ on average contains more bare and less forest land, has more widely spaced canopies, and experienced less burning during the fire. The HIZ hazard index values were spatially heterogeneous, but with several distinct high and low hazard clusters. The methods described in this study, paired with in situ data collection, can be applied to other areas to inform hazard mitigation plans.

The 20 February 2010 Madeira Island flash-floods: VHR satellite imagery processing in support of landslide inventory and sediment budget assessment

Abstract. On 20 February 2010, an extreme rainfall episode occurred on Madeira Island, which caused an exceptionally strong flash flood and several soil slip-debris flows, producing 45 confirmed deaths and 6 persons declared missing, as well as extensive material damages. In order to understand and quantify the importance of landsliding in routing sediment through mountainous drainage, such as Madeira Island's landscape, it was essential to perform extensive landslide analysis. This study describes the methodology used to semi-automatically detect the landslides, produce the landslide inventory maps and estimate the sediment volume produced during this particular event which ranged from 217 000 m3 to 344 000 m3 and 605 000 m3 to 984 000 m3 for the Funchal and Ribeira Brava basins, respectively. These results contributed to the design and implementation of measures to prevent damages caused by landslides in Madeira Island.

Information mining from human visual reasoning about multi-temporal, high-resolution satellite imagery

Over time experts in image analysis learn to use visual cues to extract large amount of information from an image in a short period of time in comparison to novice viewers. This tacit knowledge of image search strategies develops through many years of experience. Efficient interpretation appears to be a gut instinct of analysts, and this ability is difficult to verbalise or teach to the next generation of analysts. To bridge the gap between experts and novices, we propose a method to attempt to uncover visual strategies in geospatial imagery through eye tracking. Going beyond typical feature-based and classification approaches, our research fuses measures of visual attention with image-based features to derive rules through associative mining. We study how the cognitive processing of an image relates to image content. This research may open the door to a better understanding of human reasoning about the analysis of multi-temporal VHR satellite imagery.

An Assessment of Geometric Activity Features for Per-pixel Classification of Urban Man-made Objects using Very High Resolution Satellite Imagery

In this paper, the authors propose the use of Geometric Activity (GA) features for detecting man-made objects in urban areas using VHR satellite imagery. These features describe the geometric context of a pixel without the necessity of segmentation and can be integrated as extra bands in a per-pixel classification. Two main types of GA features were investigated: ridge features based on the well-known facet model and morphological features obtained by applying closing transforms with structuring elements of different size and shape. The findings show a substantial increase in classification accuracy for the man-made object classes roads and buildings with dark roof after inclusion of GA features. Next to GA features, the use of object-based features derived from eCognition®, containing both geometric and textural information, was also investigated for per-pixel classification. Accuracies obtained with object-based features are comparable to the accuracies obtained with GA features. The inclusion of both GA features and object-based features further improves the overall accuracy. GA features and object-based features thus contain complementary information.

Identification of land cover alterations in the Alta Murgia National Park (Italy) with VHR satellite imagery

Landcoverexertsagreatinfluenceonmanybasicenvironmentalprocessesandconsequentlyanytransformation in it may have a marked impact on the environment from the local to the global scales. In multidisciplinary research contexts, satellite remote sensing offers opportunities both to evaluate the effects of these processes and to provide one of the information layers needed for designing national strategies oriented to protection and sustainableuseofourresources.Theadventofrecentsatelliteimageryhasincreasedthepossibilitytoinvestigate large-scale areas in great detail. Together with an increase in spatial and radiometric resolution, there is, usually, an increase in the variability within land parcels, generating a decrease in the accuracy of land use classification on a per-pixel basis. In order to avoid such negative impacts, an object-oriented classification methodology on IKONOS multispectral data has been implemented on the test area of the Alta Murgia National Park, in the Apulia region (Italy), where soil adaptation for agricultural practices, through rock breaking, has taken place over the last 20 years. The analysis has been conducted with a classification strategy that is able to distinguish land use functions on the basis of differences in spatial distribution and pattern of land cover forms. It consists of two phases: segmentation of the image into meaningful multipixel objects of various sizes, based on both spectral and spatial characteristics of groups of pixels; then, assignment of the segments (objects) to classes using fuzzy logic and a hierarchical decision key.