normalized digital surface model (NDSM)

In 2016, DigitalGlobe's third-generation commercial Earth observation satellite WorldView-4 (WV-4), which offered 31 cm spatial resolution for panchromatic imagery, was launched and was active up to January 2019. Together with WorldView-3, this is the highest ground resolution of civilian optical space-borne missions. In this study, the information content of WV-4′s pan-sharpened stereo imagery is comprehensively investigated by manual and automatic object extraction approaches in a study area with rolling urban topography. The potential of automatic extraction was maximised by a group of refinement methods supported by the normalised digital surface model (NDSM) generated from WV-4 stereo pair. The NDSM was used as an additional band in the segmentation for revealing the object height heterogeneity. The results demonstrated that our method increased the precision, completeness and overall quality of WV-4 automatic extraction up to 8.4% for the number of buildings and up to 3.4% for the road length. Keywords WorldView-4 • Information content • Normalised digital surface model • Manual object extraction • Automatic object extraction Zusammenfassung Optimierung des erreichbaren Informationsinhalts von WorldView-4 Stereomodellen. Im Jahr 2016 startete DigitalGlobe WorldView-4 (WV-4), einen kommerziellen Erdbeobachtungssatelliten der dritten Generation. Er lieferte eine Bodenauflösung von 31 cm und war bis Januar 2019 in Betrieb. Gemeinsam mit WorldView-3 hatte er die höchste Bodenauflösung eines nicht-militärischen optischen Fernerkundungssatelliten. In dieser Arbeit wird der Informationsgehalt der pan-geschärften Stereomodelle des WV-4 umfassend durch manuelle und automatische Datenerfassung in einem Gebiet untersucht, das durch Verstädterung und bewegte Topographie geprägt ist. Das Potenzial der automatischen Datenerfassung wurde durch eine Reihe von verfeinerten Methoden maximiert, unterstützt durch das aus dem WV-4 Stereomodell erzeugte normalisierte digitale Oberflächenmodell (NDSM). Der NDSM wurde als zusätzliche Ebene in der Segmentierung verwendet, um die Heterogenität der Objekthöhen zu ermitteln. Die Ergebnisse zeigten, dass unsere Methode die Genauigkeit, Vollständigkeit und Gesamtqualität der automatischen Datenerfassung bei WV-4-Szenen bis zu 8,4% bezüglich der Anzahl der Gebäude und bis zu 3,4% bezüglich der Straßenlänge gesteigert hat.

Object-based image analysis is becoming increasingly popular in classification of very high resolution (VHR) imagery over urban areas. The spectral resolution of VHR imagery (generally they possesses 1 pan and 4 multispectral bands), however, is limited and insufficient for differentiating many urban land cover classes. Due to the spectral similarity of building roofs, roads and parking lots, spectral-based classifications which solely rely on spectral information of the image do not have promising results when applied to VHR imagery over urban landscapes. In recent years, significant amount of research has been carried out on incorporating LiDAR derived DSM into the classification to address the problems of differentiating spectrally similar objects in urban areas. However, LiDAR DSMs are expensive and not available for many urban areas. In this research, we introduce a new approach for classifying urban land cover classes by incorporating widely available photogrammetrically-derived DSMs. Even though the accuracy of photogrammetrically-derived DSMs is far below that of LiDAR DSMs, and significant misregistration exists between VHR imagery and DSM, object-based hierarchical fuzzy classification still achieve successful separation between building roofs and traffic areas. Stereo aerial photos and a pansharped QuickBird multispectral image of the downtown area of the city of Fredericton, Canada, were used for this research. Results show that buildings can be well separated from roads and parking lots, and the proposed approach has the potential to replace LiDAR DSM for urban land cover classification.

Assessment of precise, fast, and up-to-date building and building floor data are essential for urban design, planning, and management and urban environmental studies. These maps are frequently required for tracking building construction speed, monitoring horizontal and vertical urban growth and illegal constructions, updating building records, preparing reasonable urban plans, assessing hazard and risk, and generating infrastructure plans. In this study, a combination of aerial photographs and LiDAR data are used to produce individual building heights and then estimate building the floor. The two recorded laser pulse reflections of LiDAR point cloud data: First pulse (FP) and last pulse (LP) reflections were used to obtain the digital surface model (DSM) and the digital terrain model (DTM), respectively. The normalized digital surface model (nDSM) is calculated to obtain the height values of each building. The accuracy of the proposed algorithm is assessed for each building floor using ground truth data and has shown an overall accuracy of 79% and a kappa equal to 0.74 which is a promising result.

As more complete 3D information can be extracted from a stereo pair, using stereo images can provide better changes detection from LiDAR data than using shadows in a single image. However, the quality problem of 3D information in both LiDAR data and stereo images will result in false alarms and directly affect the accurate change detection. LiDAR point clouds are sparse and irregularly spaced, and have mixed returns near building edges, while 3D information extracted from stereo images are affected by shadow and low texture. This chapter proposes LiDAR-guided edge-aware dense matching (LEAD-Matching) to address these problems explicitly for detecting accurate building changes. Data sparsity and irregular spacing is addressed by densifying LiDAR points in a form of a digital surface model (DSM). Instead of applying interpolation with associated edge problems due to mixed returns, three candidate DSMs are created by linking each DSM pixel to up to three planes as identified in segmented and triangulated LiDAR data. The candidate DSMs limit the disparity search space for dense matching, addressing low texture and shadow problems in images. Through edge-aware dense matching, the detailed building edges in stereo pairs determine the optimal heights to address LiDAR edge problem. Changes are detected where corresponding pixels from dense matching have large color differences. Due to homogeneous surroundings and shadows, only partial changes are initially detected. A second hierarchical dense matching step is employed to complete changes and update 3D information by propagating initial partial changes iteratively. The proposed method is applied on data from two cities, Amersfoort and Assen, the Netherlands, with around 1200 existing buildings. In both areas, the method successfully verifies unchanged buildings while detecting minimum changes of 2 × 2 × 2 m 3. The experiments also show that the proposed method outperforms two well-known change detection methods in terms of verifying unchanged buildings and detecting small changes simultaneously.

The current practice in land cover/land use change analysis relies heavily on the individually classified maps of the multitemporal data set. Due to varying acquisition conditions (e.g., illumination, sensors, seasonal differences), the classification maps yielded are often inconsistent through time for robust statistical analysis. 3D geometric features have been shown to be stable for assessing differences across the temporal data set. Therefore, in this article we investigate the use of a multitemporal orthophoto and digital surface model derived from satellite data for spatiotemporal classification. Our approach consists of two major steps: generating per-class probability distribution maps using the random-forest classifier with limited training samples, and making spatiotemporal inferences using an iterative 3D spatiotemporal filter operating on per-class probability maps. Our experimental results demonstrate that the proposed methods can consistently improve the individual classification results by 2%-6% and thus can be an important postclassification refinement approach.

Looking at a piece of land (landscape) within a particular area as a whole, it is seen to have a heterogeneous structure peculiar to time and place. In such a structure, different developments, changes and dynamics appear depending on time and place. Landscape

In this paper, we present an analysis on the impact of network parameters for semantic segmentation architectures in context of UAV data processing. We present the analysis on the DroneDeploy Segmentation benchmark. Based on the comparative analysis we identify the optimal network architecture to be FPN-EfficientNetB3 with pretrained encoder backbones based on Imagenet Dataset. The network achieves IoU score of 0.65 and F1-score of 0.71 over the validation dataset. We also compare the various architectures in terms of their memory footprint and inference latency with further exploration of the impact of TensorRT based optimizations. We achieve memory savings of ~4.1x and latency improvement of 10% compared to Model: FPN and Backbone: InceptionResnetV2.

Bilindigi uzere; ulke yonetiminde, planlama faaliyetlerinde, tasinmazlarin vergilendirilmesinde ve mulkiyetin guvence altina alinmasinda, harita ve haritaya dayali urunler vazgecilmez altliklardir. Ulkemizde haritacilik faaliyetlerini yuruten bircok kurum bulunmaktadir. Bu kurumlar birbirinden bagimsiz calismalar yapmis olmalarina ragmen, altlik olarak kullandiklari verileri sadece Tapu Kadastro Genel Mudurlugunun ilgili birimlerinden temin etmektedirler.Yapilan calismalar kapsaminda ulkemizde haritacilik calismalarini yuruten kurumlarin kullandiklari harita altliklari incelemis, tum bu kurumlarin faaliyetlerinde kadastral altliklari baz alarak calismalarini yuruttukleri gorulmustur. Secilen uygulama sahasina ait olup grafik yontemlerde uretilmis olan kadastro paftalari sayisallastirilmis ve ulke koordinat sistemine donusturulmustur. Yapilan calismalar sonucunda mevcut arazi ile kadastronun altlik olarak kullandigi paftalar arasinda uyusumsuzluk oldugu tespit edilmistir. Bu nedenle,...

The objective of this research work is twofold: first to compare Worldview-2, IKONOS and Quickbird imagery with respect to urban green classification (the evergreen and deciduous trees) in two central urban parks in Thessaloniki, Greece, and second to obtain the best accuracy in matching the pixel-level with the object-based classifications. Representative classification algorithms, such as Maximum Likelihood classification and Decision Tree algorithm, were tested at the pixel level, and in the classification process the indices of Normalized Difference Vegetation Index (NDVI), Modified Soil Adjusted Vegetation Index (MSAVI) and Normalized Difference Water Index (NDWI) were computed on the images. The available normalized Digital Surface Model (nDSM) of the study areas improved even more the classification results. For the evaluation of the resulted classified images statistical tests were applied, and the results suggested that the developed classification system using NDVI and nDSM achieved the highest accuracy in comparison to traditional pixel-based classification methods, whereas object-oriented classification through the Decision Tree algorithm gave better accuracy results than other methods.

In remote sensing, for applications as environment monitoring, change detection based on image processing is one of the most important techniques. To reach high performance various techniques of fusion are exploited using a combination of multi-temporal, multispectral and panchromatic satellite images. A solution for handling such kind of images holds when using some simple statistical methods like the Percent Difference (PD) technique as well as the Principal Component Analysis (PCA) one. In this paper, an automatic change detection method issued from the two previous techniques is proposed and applied on multispectral and panchromatic images captured by a high resolution optical satellite. This approach is characterized by two aspects: the first one consists of the fusion of the different data and the second one performs the detection of the changes for the resulting images. The experimental results show the reasonable quantitative performance and the effectiveness of the proposed method for change detection, consisting of an automatic extraction of most of change information as well as the obtention of better results for most precision metrics consisting of an overall accuracy of up to 91% and a Kappa coefficient of up to 66%, comparing to those obtained using the simple PD and PCA techniques.

Worldview-3 stereo-extracted DSMs represent state-of-the-art products in the domain of satellite-based digital surface modelling. Main goal of our research was to evaluate the vertical accuracy of WV-3 derived DSMs over olive groves. Creation of high-accuracy WV-3 derived DSMs would allow efficient large scale management and protection of this valuable agricultural resource. Vertical accuracy of WV-3 derived DSM was evaluated at two test sites within Olive Gardens of Lun (Pag Island, Croatia), through the comparison with reference UAV photogrammetry derived VHR DSM. Two test sites were selected by object-based approach, established on spectral (NDVI, VARI) and height information (digital olive models (DOMs)). While first test site covers one single, individual oldest olive tree (45 m²), second test site covers larger area (2500 m²) with dense, unattended olive trees. Although vertical accuracy of individual olive trees still significantly deviates from reference model (RMSE = 3.604 m; MAE = 3.203 m), accuracy within larger test was much better (RMSE = 1.462 m; MAE = 1.127 m). This demonstrated that WV-3 stereo imagery has great potential for application in creation of DSMs over large scale forested areas, that would be hard to cover with field geospatial techniques (e.g. LiDAR or UAV photogrammetry).

Urban green infrastructure offers various ecosystem services that are essential for the well-being of urban residents (Pu 2021). Trees in urban areas are particularly important as they reduce water and air pollution, provide shade and cooling, reduce energy consumption, and enhance the aesthetic value of the city (Jombo et al. 2022). However, the classification of tree species in urban environments is challenging due to the complexity and variability of urban features (Alonzo et al. 2014). The accurate mapping of urban tree species is necessary for understanding their distribution and extent, which is essential for managing urban tree species and creating sustainable urban environments (Pu and Landry 2020). Remote sensing technologies offer a more efficient and cost-effective alternative to traditional methods, such as field surveys and manual interpretation of aerial maps, which can be labour-intensive, time-consuming, and expensive (Wan et al. 2021). By using remote sensing technologies, such as LiDAR and optical data, urban tree canopy and species information can be obtained quickly, easily, and repeatedly (Michałowska and Rapiński 2021). This eliminates the need for extensive manual labour and reduces the overall cost of obtaining this vital information, making it a convenient option for researchers, cartographers, urban planners, and policymakers. Combining Light Detection and Ranging (LiDAR) data with optical data has been shown to significantly enhance the accuracy of mapping urban tree species (Hartling et al. 2019). This is due to the combination of textural and spectral information extracted from the optical data, along with the three-dimensional spatial information obtained from the LiDAR data (Wallace et al. 2021). LiDAR sensors utilize high-density pulses to scan objects on the ground and record the reflected light, enabling the generation of detailed and accurate three-dimensional representations of urban environments (Kelly and Di Tommaso 2015). In this study, we employed an extreme gradient boosting (XGBoost) algorithm to map urban tree species using LiDARderived normalized Digital Surface Model (nDSM), WorldView-2 (WV-2) imagery, and vegetation indices in the city of Johannesburg. The study aimed to determine the significance of nDSM, WV-2 bands, and vegetation indices in mapping urban tree species and to compare the accuracy of using WV-2 data alone and combining the three datasets. We selected a heterogeneous urban environment for our study to demonstrate the challenges of mapping urban trees in complex urban landscapes. We obtained a WV-2 image and LiDAR data for the study area and generated a normalized Digital Surface Model (nDSM) from the LiDAR data. In order to match the spatial resolution of the WV-2 image used in this study, a Digital Terrain Model (DTM) with a 2m spatial resolution was created by utilizing the ground points. The Inverse Distance Weighting (IDW) method, as described by Xiaoye (2008), was applied to produce the DTM. The Digital Surface Model (DSM) was transformed into a raster format utilizing first return data, with a 2m cell resolution. In this research, an nDSM was created by taking the difference between the DSM and DTM. It was assumed that each tree crown had an area of 3m², and the minimum height of a tree was 3m from the ground. To remove small features such as shrubs and grasslands from the urban tree species class, the study employed the values for crown size and height. The nDSM layer was utilized to display the highest points of reflective surfaces that were assumed to be tree canopies or building tops. Vegetation indices were used to differentiate tree canopies and building tops. We extracted vegetation indices from the WV-2 image and computed texture features from the nDSM. We then trained an XGBoost model to classify the tree species based on the input features. We evaluated the performance of the XGBoost model using a confusion matrix and calculated the overall accuracy, kappa coefficient, and F1 score. Our results highlight that the accuracy of mapping urban trees can be significantly improved by combining LiDAR and optical data. We achieved an overall accuracy of 89% using the WV-2 data alone and 95% on the combined dataset. The nDSM was the topranked variable with an F Score of 0.86, followed by the red edge vegetation index with an F Score of 0.70. The study provides valuable insights into the mapping of urban tree species using LiDAR and optical data. Our findings demonstrate that the combined dataset of LiDAR-derived nDSM, WV-2 imagery, and vegetation indices can significantly improve the accuracy of mapping urban trees in heterogeneous urban environments. The study highlights

Slums, characterized by substandard housing conditions, are a common in fast growing Asian cities. However, reliable and up-to-date information on their locations and development dynamics is scarce. Despite numerous studies, the task of delineating slum areas remains a challenge and no general agreement exists about the most suitable method for detecting or assessing detection performance. In this paper, standard computer vision methods-Bag of Visual Words framework and Speeded-Up Robust Features have been applied for image-based classification of slum and non-slum areas in Kalyan and Bangalore, India, using very high resolution RGB images. To delineate slum areas, image segmentation is performed as pixel-level classification for three classes: Slums, Built-up and Non-Built-up. For each of the three classes, image tiles were randomly selected using ground truth observations. A multi-class support vector machine classifier has been trained on 80% of the tiles and the remaining 20% were used for testing. The final image segmentation has been obtained by classification of every 10th pixel followed by a majority filtering assigning classes to all remaining pixels. The results demonstrate the ability of the method to map slums with very different visual characteristics in two very different Indian cities.

Migratory waterfowl (i.e., ducks, geese, and swans) management relies on landscape bioenergetic models to inform on-the-ground habitat conditions and conservation practices. Therefore, conservation planners rely on accurate predictions of wetland habitats for waterfowl at regional scales. Unharvested flooded corn is a popular management tool on public and private lands that greatly increases landscape-level energy compared to other wetlands; thus, landscape bioenergetic models are particularly sensitive to these habitat features. Despite their importance to conservation planning and implementation, the abundance and distribution of unharvested flooded corn fields across North America is unknown. Furthermore, training data is difficult to collect and accurate predictions are challenging given their unique attributes and discreteness at landscape-level lens. Advances in multispectral imagery and deep learning algorithms may enable continuous and autonomous detection of these habitat f...

Migratory waterfowl (i.e., ducks, geese, and swans) management relies on landscape bioenergetic models to inform on-the-ground habitat conditions and conservation practices. Therefore, conservation planners rely on accurate predictions of wetland habitats for waterfowl at regional scales. Unharvested flooded corn is a popular management tool on public and private lands that greatly increases landscape-level energy compared to other wetlands; thus, landscape bioenergetic models are particularly sensitive to these habitat features. Despite their importance to conservation planning and implementation, the abundance and distribution of unharvested flooded corn fields across North America is unknown. Furthermore, training data is difficult to collect and accurate predictions are challenging given their unique attributes and discreteness at landscape-level lens. Advances in multispectral imagery and deep learning algorithms may enable continuous and autonomous detection of these habitat f...

The latest breed of very high resolution (VHR) commercial satellites opens new possibilities for cartographic and remote-sensing applications. In fact, one of the most common applications of remote-sensing images is the extraction of land-cover information for digital image base maps by means of classification techniques. The aim of the study was to compare the potential classification accuracy provided by pan-sharpened orthoimages from both GeoEye-1 and WorldView-2 (WV2) VHR satellites over urban environments. The influence on the supervised classification accuracy was evaluated by means of an object-based statistical analysis regarding three main factors: (i) sensor used; (ii) sets of image object (IO) features used for classification considering spectral, geometry, texture, and elevation features; and (iii) size of training samples to feed the classifier (nearest neighbour (NN)). The new spectral bands of WV2 (Coastal, Yellow, Red Edge, and Near Infrared-2) did not improve the benchmark established from GeoEye-1. The best overall accuracy for GeoEye-1 (close to 89%) was attained by using together spectral and elevation features, whereas the highest overall accuracy for WV2 (83%) was achieved by adding textural features to the previous ones. In the case of buildings classification, the normalized digital surface model computed from light detection and ranging data was the most valuable feature, achieving producer's and user's accuracies close to 95% and 91% for GeoEye-1 and VW2, respectively. Last but not least and regarding the size of the training samples, the rule of 'the larger the better' was true but, based on statistical analysis, the ideal choice would be variable depending on both each satellite and target class. In short, 20 training IOs per class would be enough if the NN classifier was applied on pan-sharpened orthoimages from both GeoEye-1 and WV2.

In remote sensing, for applications as environment monitoring, change detection based on image processing is one of the most important techniques. To reach high performance various techniques of fusion are exploited using a combination of multi-temporal, multispectral and panchromatic satellite images. A solution for handling such kind of images holds when using some simple statistical methods like the Percent Difference (PD) technique as well as the Principal Component Analysis (PCA) one. In this paper, an automatic change detection method issued from the two previous techniques is proposed and applied on multispectral and panchromatic images captured by a high resolution optical satellite. This approach is characterized by two aspects: the first one consists of the fusion of the different data and the second one performs the detection of the changes for the resulting images. The experimental results show the reasonable quantitative performance and the effectiveness of the proposed method for change detection, consisting of an automatic extraction of most of change information as well as the obtention of better results for most precision metrics consisting of an overall accuracy of up to 91% and a Kappa coefficient of up to 66%, comparing to those obtained using the simple PD and PCA techniques.

Building extraction from high resolution (HR) satellite imagery is one of the most significant issue for remote sensing community. Manual extraction process is onerous and time consuming that's why the improvement of the best automation is a crucial topic for the researchers. In this study, we aimed to expose the significant contribution of normalized digital surface model (nDSM) to the automatic building extraction from mono HR satellite imagery performing two-step application in an appropriate study area which includes various terrain formations. In first step, the buildings were manually and object-based automatically extracted from ortho-rectified pan-sharpened IKONOS and Quickbird HR imagery that have 1 m and 0.6 m ground sampling distances (GSD), respectively. Next, the nDSM was created using available aerial photos to represent the height of individual non-terrain objects and used as an additional channel for segmentation. All of the results were compared with the reference data, produced from aerial photos that have 5 cm GSD. With the contribution of nDSM, the number of extracted buildings was increased and more importantly, the number of falsely extracted buildings occurred by automatic extraction errors was sharply decreased, both are the main components of precision, completeness and overall quality.

Looking at a piece of land (landscape) within a particular area as a whole, it is seen to have a heterogeneous structure peculiar to time and place. In such a structure, different developments, changes and dynamics appear depending on time and place. Landscape

In 2016, DigitalGlobe's third-generation commercial Earth observation satellite WorldView-4 (WV-4), which offered 31 cm spatial resolution for panchromatic imagery, was launched and was active up to January 2019. Together with WorldView-3, this is the highest ground resolution of civilian optical space-borne missions. In this study, the information content of WV-4′s pan-sharpened stereo imagery is comprehensively investigated by manual and automatic object extraction approaches in a study area with rolling urban topography. The potential of automatic extraction was maximised by a group of refinement methods supported by the normalised digital surface model (NDSM) generated from WV-4 stereo pair. The NDSM was used as an additional band in the segmentation for revealing the object height heterogeneity. The results demonstrated that our method increased the precision, completeness and overall quality of WV-4 automatic extraction up to 8.4% for the number of buildings and up to 3.4% for the road length. Keywords WorldView-4 • Information content • Normalised digital surface model • Manual object extraction • Automatic object extraction Zusammenfassung Optimierung des erreichbaren Informationsinhalts von WorldView-4 Stereomodellen. Im Jahr 2016 startete DigitalGlobe WorldView-4 (WV-4), einen kommerziellen Erdbeobachtungssatelliten der dritten Generation. Er lieferte eine Bodenauflösung von 31 cm und war bis Januar 2019 in Betrieb. Gemeinsam mit WorldView-3 hatte er die höchste Bodenauflösung eines nicht-militärischen optischen Fernerkundungssatelliten. In dieser Arbeit wird der Informationsgehalt der pan-geschärften Stereomodelle des WV-4 umfassend durch manuelle und automatische Datenerfassung in einem Gebiet untersucht, das durch Verstädterung und bewegte Topographie geprägt ist. Das Potenzial der automatischen Datenerfassung wurde durch eine Reihe von verfeinerten Methoden maximiert, unterstützt durch das aus dem WV-4 Stereomodell erzeugte normalisierte digitale Oberflächenmodell (NDSM). Der NDSM wurde als zusätzliche Ebene in der Segmentierung verwendet, um die Heterogenität der Objekthöhen zu ermitteln. Die Ergebnisse zeigten, dass unsere Methode die Genauigkeit, Vollständigkeit und Gesamtqualität der automatischen Datenerfassung bei WV-4-Szenen bis zu 8,4% bezüglich der Anzahl der Gebäude und bis zu 3,4% bezüglich der Straßenlänge gesteigert hat.

Building extraction from high resolution (HR) satellite imagery is one of the most significant issue for remote sensing community. Manual extraction process is onerous and time consuming that's why the improvement of the best automation is a crucial topic for the researchers. In this study, we aimed to expose the significant contribution of normalized digital surface model (nDSM) to the automatic building extraction from mono HR satellite imagery performing two-step application in an appropriate study area which includes various terrain formations. In first step, the buildings were manually and object-based automatically extracted from ortho-rectified pan-sharpened IKONOS and Quickbird HR imagery that have 1 m and 0.6 m ground sampling distances (GSD), respectively. Next, the nDSM was created using available aerial photos to represent the height of individual non-terrain objects and used as an additional channel for segmentation. All of the results were compared with the reference data, produced from aerial photos that have 5 cm GSD. With the contribution of nDSM, the number of extracted buildings was increased and more importantly, the number of falsely extracted buildings occurred by automatic extraction errors was sharply decreased, both are the main components of precision, completeness and overall quality.

Building extraction from high resolution (HR) satellite imagery is one of the most significant issue for remote sensing community. Manual extraction process is onerous and time consuming that's why the improvement of the best automation is a crucial topic for the researchers. In this study, we aimed to expose the significant contribution of normalized digital surface model (nDSM) to the automatic building extraction from mono HR satellite imagery performing two-step application in an appropriate study area which includes various terrain formations. In first step, the buildings were manually and object-based automatically extracted from ortho-rectified pan-sharpened IKONOS and Quickbird HR imagery that have 1 m and 0.6 m ground sampling distances (GSD), respectively. Next, the nDSM was created using available aerial photos to represent the height of individual non-terrain objects and used as an additional channel for segmentation. All of the results were compared with the reference data, produced from aerial photos that have 5 cm GSD. With the contribution of nDSM, the number of extracted buildings was increased and more importantly, the number of falsely extracted buildings occurred by automatic extraction errors was sharply decreased, both are the main components of precision, completeness and overall quality.

In remote sensing, for applications as environment monitoring, change detection based on image processing is one of the most important techniques. To reach high performance various techniques of fusion are exploited using a combination of multi-temporal, multispectral and panchromatic satellite images. A solution for handling such kind of images holds when using some simple statistical methods like the Percent Difference (PD) technique as well as the Principal Component Analysis (PCA) one. In this paper, an automatic change detection method issued from the two previous techniques is proposed and applied on multispectral and panchromatic images captured by a high resolution optical satellite. This approach is characterized by two aspects: the first one consists of the fusion of the different data and the second one performs the detection of the changes for the resulting images. The experimental results show the reasonable quantitative performance and the effectiveness of the proposed method for change detection, consisting of an automatic extraction of most of change information as well as the obtention of better results for most precision metrics consisting of an overall accuracy of up to 91% and a Kappa coefficient of up to 66%, comparing to those obtained using the simple PD and PCA techniques.

Biogeosciences and Forestry Biogeosciences and Forestry Three-dimensional forest stand height map production utilizing airborne laser scanning dense point clouds and precise quality evaluation Umut G Sefercik (1) , Ayhan Atesoglu (2) In remote sensing, estimation of the forest stand height is an ever-challenging issue due to the difficulties encountered during the acquisition of data under forest canopies. Stereo optical imaging offers high spatial and spectral resolution; however, the optical correlation is lower in dense forests than in open areas due to an insufficient number of matching points. Therefore, in most cases height information may be missing or faulty. With their long wavelengths of 0.2 to 1.3 m, P-band and L-band synthetic aperture radars are capable of penetrating forest canopies, but their low spatial resolutions restrict the use of single-tree based forest applications. In this study, airborne laser scanning was used as an effective remote sensing technique to produce large-scale maps of forest stand height. This technique produces very high-resolution point clouds and has a high penetration capability that allows for the detection of multiple echoes per laser pulse. A study area with a forest coverage of approximately 60% was selected in Houston, USA, and a three-dimensional colorcoded map of forest stands was produced using a normalized digital surface model technique. Rather than being limited to the number of ground control points, the accuracy of the produced map was assessed with a model-to-model approach using terrestrial laser scanning. In the accuracy assessment, the standard deviation was used as the main accuracy indicator in addition to the root mean square error and normalized median absolute deviation. The absolute geo-location accuracy of the generated map was found to be better than 1 cm horizontally and approximately 40 cm in height. Furthermore, the effects of bias and relative standard deviations were determined. The problems encountered during the production of the map, as well as recommended solutions, are also discussed in this paper.

Looking at a piece of land (landscape) within a particular area as a whole, it is seen to have a heterogeneous structure peculiar to time and place. In such a structure, different developments, changes and dynamics appear depending on time and place. Landscape

Received: 12 June 2020 Accepted: 23 December 2020 Urban building information can be effectively extracted by applying object-based image segmentation and multi-stage thresholding on High Resolution (HR) remote sensing satellite imageries. This study provides the results obtained using this method on the images of Indian remote sensing satellite, CARTOSAT-2S launched by the Indian Space Research Organization (ISRO). In this study, a method is developed to extract urban building footprints from the HR remote sensing satellite images. The first step of the process consists of generating highly dense per pixel Digital Surface Model (DSM) by using semi global matching algorithm on HR satellite stereo images and applying robust ground filtering to generate Digital Terrain Model (DTM). In the second step, multi-stage object-based approach is adopted to extract building bases using the PAN sharpened image, normalized Digital Surface Model (nDSM) derived from DSM and DTM, and Normalised Diff...

Yasadışı yapılaşma, çarpık kentleşmeye ve dengesiz kent yapısına neden olmakta, özellikle yüksek nüfuslu kentlerde teknik ve sosyal açılardan büyük bir sorun oluşturabilmektedir. İmar uygulaması yapılacak kentsel alanlar hakkında detaylı teknik bilgiler veren imar planları, bir binanın zemin üzerinde kaplaması gereken maksimum alanı ve kat adetleri ile oranlarının sınır değerlerini de vermektedir. Bu planlar, her ne kadar olması gereken kent yapısı için ideali belirtse de pratikte uygulanması, özellikle İstanbul gibi kentsel dönüşüm geçiren kentlerde zorlayıcı olabilmektedir. Bu çalışmada, İstanbul'un Üsküdar ilçesinde Güzeltepe mahallesine ait hava LIDAR nokta bulutu verileri ile bölgenin imar planları karşılaştırılmış ve bu verilerin planlarla ne kadar uyuştuğu araştırılmıştır. Mevcut imar planları CBS ortamına aktarılmış ve LIDAR verilerinden, dijital yüzey modeli (DSM), dijital arazi modeli (DTM) ve normalize edilmiş dijital yüzey modeli (nDSM) üretilerek bina yükseklikleri elde edilmiştir. Ayrıca bina verileri CityGML semantik yapısına dönüştürülerek binalara ait çatı ve cephe detaylarının ayrımı da yapılmıştır. Bu bina yükseklik verileri imar planlarındaki sınır değerler ile karşılaştırılmıştır. Bu çalışmadan elde edilen sonuçların, otomatik bina çıkarımı ve 3D model üretimi ile kentsel değişim kontrolünde uygulanma potansiyeli olabilir.

The automatic extraction of information content from remotely sensed data is always challenging. We suggest a novel fusion approach to improve the extraction of this information from mono-satellite images. A Worldview-2 (WV-2) pan-sharpened image and a 1/5000-scaled topographic vector map (TOPO5000) were used as the sample data. Firstly, the buildings and roads were manually extracted from WV-2 to point out the maximum extractable information content. Subsequently, object-based automatic extractions were performed. After achieving two-dimensional results, a normalized digital surface model (nDSM) was generated from the underlying digital aerial photos of TOPO5000, and the automatic extraction was repeated by fusion with the nDSM to include individual object heights as an additional band for classification. The contribution was tested by precision, completeness and overall quality. Novel fusion technique increased the success of automatic extraction by 7% for the number of buildings and by 23% for the length of roads.

In remote sensing, estimation of the forest stand height is an ever-challenging issue due to the difficulties encountered during the acquisition of data under forest canopies. Stereo optical imaging offers high spatial and spectral resolution; however, the optical correlation is lower in dense forests than in open areas due to an insufficient number of matching points. Therefore, in most cases height information may be missing or faulty. With their long wavelengths of 0.2 to 1.3 m, P-band and L-band synthetic aperture radars are capable of penetrating forest canopies, but their low spatial resolutions restrict the use of single-tree based forest applications. In this study, airborne laser scanning was used as an effective remote sensing technique to produce large-scale maps of forest stand height. This technique produces very high-resolution point clouds and has a high penetration capability that allows for the detection of multiple echoes per laser pulse. A study area with a forest coverage of approximately 60% was selected in Houston, USA, and a three-dimensional colorcoded map of forest stands was produced using a normalized digital surface model technique. Rather than being limited to the number of ground control points, the accuracy of the produced map was assessed with a model-to-model approach using terrestrial laser scanning. In the accuracy assessment, the standard deviation was used as the main accuracy indicator in addition to the root mean square error and normalized median absolute deviation. The absolute geo-location accuracy of the generated map was found to be better than 1 cm horizontally and approximately 40 cm in height. Furthermore, the effects of bias and relative standard deviations were determined. The problems encountered during the production of the map, as well as recommended solutions, are also discussed in this paper.

Three-dimensional urban cartography is needed for city changes' assessment. The variety of studies using 3D calculations of urban elements grows each year. Building and vegetation volumes are necessary to assess and understand spatio-temporal urban changeable environments. However, there are technical questions as to which method can improve 3D urban cartographic accuracy. The innovative part of this current study is the creation of a six-band hybrid obtained from LIDAR and WorldView2 synergy. Two different enhancement algorithms demonstrated the most important spectral features for the urban development and vegetation classes. Results indicated an improvement in accuracy by up to 21.3%, according to the Kappa coefficient. Both infra-red band and intensity band were the most significant, according to the principal components analysis. The synergy delimited classes and polygons, as well as the direct display of information regarding heights of elements and improving the extraction of roads, buildings and vegetation classes.

Up to date building information is important for urban planning and urban management applications. Since the very high resolution (VHR) remotely sensed images are the most cost effective geo-data for mapping building information, building detection in VHR images has become an extremely attractive area of research. Buildings are inherently elevated objects. Thus, detection procedures that are supported by elevation information provide more reliable building delineation results. However, all existing elevationbased building detection methods need first to generate a normalized digital surface model (nDSM) to remove the influence of terrain elevation. An nDSM is typically calculated by identifying the digital terrain model (DTM) in a digital surface model (DSM) and then subtracting the DTM from the DSM. Although nDSMs are commonly used in building detection, the generation of an accurate nDSM is still a challenging task in many urban areas. Therefore, we introduce in this paper a slope...

The automatic extraction of information content from remotely sensed data is always challenging. We suggest a novel fusion approach to improve the extraction of this information from mono-satellite images. A Worldview-2 (WV-2) pan-sharpened image and a 1/5000-scaled topographic vector map (TOPO5000) were used as the sample data. Firstly, the buildings and roads were manually extracted from WV-2 to point out the maximum extractable information content. Subsequently, object-based automatic extractions were performed. After achieving two-dimensional results, a normalized digital surface model (nDSM) was generated from the underlying digital aerial photos of TOPO5000, and the automatic extraction was repeated by fusion with the nDSM to include individual object heights as an additional band for classification. The contribution was tested by precision, completeness and overall quality. Novel fusion technique increased the success of automatic extraction by 7% for the number of buildings and by 23% for the length of roads.

Up to date building information is important for urban planning and urban management applications. Since the very high resolution (VHR) remotely sensed images are the most cost effective geo-data for mapping building information, building detection in VHR images has become an extremely attractive area of research. Buildings are inherently elevated objects. Thus, detection procedures that are supported by elevation information provide more reliable building delineation results. However, all existing elevationbased building detection methods need first to generate a normalized digital surface model (nDSM) to remove the influence of terrain elevation. An nDSM is typically calculated by identifying the digital terrain model (DTM) in a digital surface model (DSM) and then subtracting the DTM from the DSM. Although nDSMs are commonly used in building detection, the generation of an accurate nDSM is still a challenging task in many urban areas. Therefore, we introduce in this paper a slope...

Knowledge of surface albedo at individual roof scale is important for mitigating urban heat islands and understanding urban climate change. This study presents a method for quantifying surface albedo of individual roofs in a complex urban area using the integration of Landsat 8 and airborne LiDAR data. First, individual roofs were extracted from airborne LiDAR data and orthophotos using optimized segmentation and supervised object based image analysis (OBIA). Support vector machine (SVM) was used as a classifier in OBIA process for extracting individual roofs. The user-defined parameters required in SVM classifier were selected using v-fold cross validation method. After that, surface albedo was calculated for each individual roof from Landsat images. Finally, thematic maps of mean surface albedo of individual roofs were generated in GIS and the results were discussed. Results showed that the study area is covered by 35% of buildings varying in roofing material types and conditions....

Biogeosciences and Forestry Biogeosciences and Forestry Three-dimensional forest stand height map production utilizing airborne laser scanning dense point clouds and precise quality evaluation Umut G Sefercik (1) , Ayhan Atesoglu (2) In remote sensing, estimation of the forest stand height is an ever-challenging issue due to the difficulties encountered during the acquisition of data under forest canopies. Stereo optical imaging offers high spatial and spectral resolution; however, the optical correlation is lower in dense forests than in open areas due to an insufficient number of matching points. Therefore, in most cases height information may be missing or faulty. With their long wavelengths of 0.2 to 1.3 m, P-band and L-band synthetic aperture radars are capable of penetrating forest canopies, but their low spatial resolutions restrict the use of single-tree based forest applications. In this study, airborne laser scanning was used as an effective remote sensing technique to produce large-scale maps of forest stand height. This technique produces very high-resolution point clouds and has a high penetration capability that allows for the detection of multiple echoes per laser pulse. A study area with a forest coverage of approximately 60% was selected in Houston, USA, and a three-dimensional colorcoded map of forest stands was produced using a normalized digital surface model technique. Rather than being limited to the number of ground control points, the accuracy of the produced map was assessed with a model-to-model approach using terrestrial laser scanning. In the accuracy assessment, the standard deviation was used as the main accuracy indicator in addition to the root mean square error and normalized median absolute deviation. The absolute geo-location accuracy of the generated map was found to be better than 1 cm horizontally and approximately 40 cm in height. Furthermore, the effects of bias and relative standard deviations were determined. The problems encountered during the production of the map, as well as recommended solutions, are also discussed in this paper.

The automated classification of urban areas in one of the main topic in the Geomatics domain. Several papers dealing with this topic have been already presented in the last decade. Most of these approaches uses multi-spectral or LiDAR data or both of them as input. In this paper, an algorithm for urban areas classification based only on overlapping RGB images is presented. The integration of radiometric and geometric information derived from aerial images is exploited in order to extract the three main classes of urban areas (i.e. building, vegetation and road) in automated way and without prior information. A photogrammetric Digital Surface Model (DSM) is firstly generated applying dense image matching techniques and this information as well as some spatial features provided by morphological filters are combined to derive a first classification. Subsequently, a thematic classification of the surveyed areas is performed considering the surface's reflectance in the visible spectrum of the used images and the multi-image information provided by the overlapping images. Range and image information are so merged in an algorithm that allows the reciprocal and iterative sharing of information in order to increase the reliability and completeness of the classification process. After a detailed description of the algorithm, the achieved results over dense urban areas are shown and discussed.

Fast and efficient detection and reconstruction of buildings have become essential in real-time applications such as navigation, 3D rendering, augmented reality, and 3D smart cities. In this study, a modern Deep Learning (DL)-based framework is proposed for automatic detection, localization, and height estimation of buildings, simultaneously, from a single aerial image. The proposed framework is based on a Y-shaped Convolutional Neural Network (Y-Net) which includes one encoder and two decoders. The input of the network is a single RGB image, while the outputs are predicted height information of buildings as well as the rooflines in three classes of eave, ridge, and hip lines. The extracted knowledge by the Y-Net (i.e. buildings' heights and rooflines) is utilized for 3D reconstruction of buildings based on the third Level of Detail (LoD2). The main steps of the proposed approach are data preparation, CNNs training, and 3D reconstruction. For the experimental investigations airborne data from Potsdam are used, which were provided by ISPRS. For the predicted heights, the results show an average Root Mean Square Error (RMSE) and a Normalized Median Absolute Deviation (NMAD) of about 3.8 m and 1.3 m, respectively. Moreover, the overall accuracy of the extracted rooflines is about 86%.

Nowadays, remotely sensed images are used for various purposes in different applications. One of them is the change detection using high resolution satellite imagery which is city areas. Urbanization rapid land use and land cover change has taken place in many cities of Turkey as well as in the world in the last 50 years. In this context, comparison of extraction results from these images and existing vector data is the most important issue. The goal here is to show the advantages and disadvantages of the two IKONOS images for creating road and building database and also updating to the database. In this study, high resolution IKONOS 2002 and IKONOS 2008 have been chosen for the test area of the Zonguldak city. Firstly, pan-sharpened IKONOS images have been produced by fusion of high resolution PAN and MS images using PCI Geomatica v9.1 software package. The parcel, building and road network objects from these datasets have been extracted automatically by initially dividing it into the segments and then, they have been classified by using the spectral, spatial and contextual information using eCognition v4.0.6 software package. On the other hand, these objects have been manually digitized from high resolution images using ArcGIS v9.3 software package. These vectors produced automatically and manually have been compared with the existing digital cadastral maps and reference vector maps of scale 1/5000 of test area. The success of object-oriented image analysis results was tested by GIS software; the results have been presented and commented. Therefore, making GIS-based analysis and comparisons with raster and vector data of the test area has crucial importance in terms of putting forth the recent situation.

The rapid population growth in the world and the adverse effects of climate change, has caused a severe increase in water demand used in sectors such as agriculture, industry, and domestic usage. The fact that the water resources in the world are stable and their availability is limited, the importance of water resources, efficient use of water, and the use of technologies related to alternative water resources have become important. This study aimed to investigate the usage potential of rainwater to be collected from the roofs of the building in irrigation of green areas in Mersin University -Çiftlikköy campus. For this purpose, first, whether the roofs inside the campus can collect rainwater will be revealed with unmanned aerial vehicles. Then, appropriate roof areas where rainwater will be collected will be calculated using the objectbased classification method. The rainwater amounts to be collected from each building will be calculated using the monthly precipitation data obtained from the Provincial Directorate of State Meteorology Affairs. In addition, by calculating the number of green areas on the campus and the need for rainwater, it was determined how much rainwater to be collected meets the water need according to different irrigation periods.

In 2016, DigitalGlobe’s third-generation commercial Earth observation satellite WorldView-4 (WV-4), which offered 31 cm spatial resolution for panchromatic imagery, was launched and was active up to January 2019. Together with WorldView-3, this is the highest ground resolution of civilian optical space-borne missions. In this study, the information content of WV-4′s pan-sharpened stereo imagery is comprehensively investigated by manual and automatic object extraction approaches in a study area with rolling urban topography. The potential of automatic extraction was maximised by a group of refinement methods supported by the normalised digital surface model (NDSM) generated from WV-4 stereo pair. The NDSM was used as an additional band in the segmentation for revealing the object height heterogeneity. The results demonstrated that our method increased the precision, completeness and overall quality of WV-4 automatic extraction up to 8.4% for the number of buildings and up to 3.4% for the road length.

The automatic extraction of information content from remotely sensed data is always challenging. We suggest a novel fusion approach to improve the extraction of this information from mono-satellite images. A Worldview-2 (WV-2) pan-sharpened image and a 1/5000-scaled topographic vector map (TOPO5000) were used as the sample data. Firstly, the buildings and roads were manually extracted from WV-2 to point out the maximum extractable information content. Subsequently, object-based automatic extractions were performed. After achieving two-dimensional results, a normalized digital surface model (nDSM) was generated from the underlying digital aerial photos of TOPO5000, and the automatic extraction was repeated by fusion with the nDSM to include individual object heights as an additional band for classification. The contribution was tested by precision, completeness and overall quality. Novel fusion technique increased the success of automatic extraction by 7% for the number of buildings and by 23% for the length of roads.

In remote sensing, estimation of the forest stand height is an ever-challenging issue due to the difficulties encountered during the acquisition of data under forest canopies. Stereo optical imaging offers high spatial and spectral resolution ; however, the optical correlation is lower in dense forests than in open areas due to an insufficient number of matching points. Therefore, in most cases height information may be missing or faulty. With their long wavelengths of 0.2 to 1.3 m, P-band and L-band synthetic aperture radars are capable of penetrating forest canopies, but their low spatial resolutions restrict the use of single-tree based forest applications. In this study, airborne laser scanning was used as an effective remote sensing technique to produce large-scale maps of forest stand height. This technique produces very high-resolution point clouds and has a high penetration capability that allows for the detection of multiple echoes per laser pulse. A study area with a forest coverage of approximately 60% was selected in Houston, USA, and a three-dimensional color-coded map of forest stands was produced using a normalized digital surface model technique. Rather than being limited to the number of ground control points, the accuracy of the produced map was assessed with a model-to-model approach using terrestrial laser scanning. In the accuracy assessment, the standard deviation was used as the main accuracy indicator in addition to the root mean square error and normalized median absolute deviation. The absolute geo-location accuracy of the generated map was found to be better than 1 cm horizontally and approximately 40 cm in height. Furthermore, the effects of bias and relative standard deviations were determined. The problems encountered during the production of the map, as well as recommended solutions , are also discussed in this paper.

Correct identification of obstacles at the periphery of airports is an important issue to ensure safe takeoff, flight, and landing to aircrafts. This work is carried on as part of the obstacle risk assessment and risk mitigation operations in the aviation security framework. This paper presents a novel fully automatic remote sensing methodology for the detection, shape and signature extraction and classification of obstacles to air navigation from very high resolution (VHR) multispectral (MS) satellite stereo couples images, here defined feature extraction (FE). In order to reduce the costs, the proposed technique is applied only on detailed areas where orographic/topographic changes potentially associated with variations in the obstacles to air navigation in wide areas have been previously detected through a lowcost pre-screening change detection (CD) methodology applied to cheaper high resolution (HR) satellite imagery. The combination of CD and FE strategies offers a low-cost and fast solution to the problem of updating airport obstacle chart.