جستجوی مقالات مرتبط با کلیدواژه « building extraction » در نشریات گروه « عمران »

Floods are one of the most frequent natural disasters that affect the society, impact human lives, and make costs for governments. Utilizing new technologies helps managers and first responders to decrease the damaging effect of floods and save time. Unmanned Aerial Vehicles equipped with accurate sensors along with powerful computer vision and deep learning techniques can act as potential platforms for surrveilance, mapping and detection of flooded regions. In this study, PSPNet as the main architecture enhanced by ResNeSt as the encoder, are utilized for semantic segmentation of very high resolution drone imagery acquired from urban flooded regions. Furthermore, in order to interpret and study the performance of the method, Monte-Carlo Dropout (MCD) technique is used as a Bayesian estimator for uncertainty quantification of the results. Comparing the results of our method with other models indicated that increasing the complexity and number of parameters of the model would increase its performance during training and testing by 10% and 3%, respectively, and the certainty of the models will increase in inference time. The Accuracy of semantic segmentation is 97.93% and F1-score is about 89%.

Building extraction is one of the most crucial requirements of urban planning. Due to their availability and affordable cast, high-resolution remotely sensed images are often used for building extraction. Owing to their impressive performances, Deep learning techniques have attracted the attention of researchers for building extraction from high-resolution images. Nevertheless, most existing models perform poorly in recovering spatial details and discriminating buildings with various sizes and shapes. Hence, this paper proposes an improvement module to address the problems associated with multi-scale building extraction. The proposed module uses dilated convolutions to increase the receiving information area to reduce the discontinuities in the results of large buildings. Extracting large buildings using the proposed module and small buildings using the main architecture of the network has turned the proposed network into an effective method for building extraction. The results of the experiments showed that the proposed module with the IoU of 0.6495 and 0.8572 for Massachusetts and WHU data sets outperformed FCN, U-Net, USSP, and DeepLab V3+. The performance analysis of the proposed module also showed that this module was able to improve the performance of building extraction considering the IoU metric by 0.1077.

Building extraction from different data resources, which can be applied in 3D city modeling, is one of the most growing research topics in Photogrammetry and Remote Sensing. The aim of this study is proposing a new method for extracting buildings from LiDAR point clouds using a fuzzy inference system. In the proposed method, first, a pre-processing step is performed to eliminate low altitude points and also noisy points. In this regard, an optimal noise removal technique is used to remove noise and outliers from LiDAR point clouds. The proposed method can almost eliminate all of the noises and outliers from the point clouds. After pre-processing step, only high altitude points in the form of points groups including buildings and trees will remain. Then, due to the uncertainty in distinguishing building groups from tree groups, a fuzzy inference system is designed and implemented. In the proposed fuzzy system, three geometric descriptors of "SumD", "Area" and "Volume" are considered as input variables, and feature type as output variable is defined. In order to evaluate the proposed method, a test area of Belgium is used, and the obtained results proved the ability of the proposed fuzzy inference system in resolving the uncertainty in detecting building points group. Moreover, the proposed noise removal method increased 10% in accuracy and improving the quality of the fuzzy system.
3D city Modeling using the technology of LiDAR is one of the most growing research topics in Photogrammetry and Remote Sensing.

In recent years, the techniques of surface representation have been changed and three-dimensional models have been replaced with two-dimensional maps. Airborne laser scanner as a powerful system has been known in remote sensing as a valuable source for 3D data acquisition from the Earth’s surface which can mainly be employed for 3D reconstruction and modeling. 3D reconstruction of buildings as an important element of 3D city models, based on LiDAR point clouds has been considered in this study. A new non-parametric method is proposed for generation of 3D model of buildings with flat and tilted roof. The approach comprises of four steps for 3D building reconstruction as: (A) Simultaneous building extraction and segmentation, (B) Edge detection, (C) Line approximation, and (D) 3D modeling. In step (A) a multi-agent method is proposed for extraction of buildings from LiDAR point clouds and segmentation of roof points at the same time. In this method five criteria such as height values, number of returned pulses, length, normal vector direction based on Constrained Delaunay Triangulation, and area are utilized. Next, in step (B) the edge points of roof segments are detected. Points of triangles having no neighboring triangles are extracted as primary edge points. In the extraction process, noises, external objects, and tree points on the roofs are eliminated. It is an advantage of the proposed method, however it leads to create the undesired edge points. There is the same problem regarding to segments which contain overlap with each other (like flat building). These undesired edge points as internal points are known and must be removed. In this paper, a method named “Grid Erosion” is employed for removing these internal points and therefore finding real edge points. After detecting the final edge points, a RANSAC-based algorithm is employed to approximate building lines in step (C). RANSAC is a powerful technique in line fitting and in comparison with general least square method, especially with noisy data, it provides robust results. In order to reduce the sensitivity of RANSAC to select parameters and no need for heavy post-processing, edge points are grouped by considering the angle between two consecutive connecting convex points. After classification of edge points, a RANSAC algorithm is separately applied on each classified edge-points group to produce primary lines. The regularization constraints should be applied on primary lines to generate the final lines. Finally, by modeling of the roofs and walls, 3D buildings model is reconstructed in step (D). The proposed method has been applied on the LiDAR data over the Vaihingen city, Germany. Building roof model is manually digitized from LiDAR point clouds and compared with building roof models that reconstructed using proposed method. In model reconstruction, the dominant errors are close to 30 cm which is calculated in horizontal distance. The main advantage of this method is its capability for segmentation and reconstruction of flat buildings containing parallel roof structures even with very small height differences (e.g. 10 cm). The results of both visual and quantitative assessments indicate that the proposed method could successfully extract the buildings from LiDAR data and generate the building models.

Nowadays, automatic processing and object extraction from data acquired by airborne sensors has been an important research topic in photogrammetric institutes. Airborne laser scanning system, which is commonly referred to as LiDAR, is a superior technology for three-dimensional spatial data acquisition from Earth’s surface in high speed and density. 3D city modeling is one of the main applications of LiDAR system. An important first step to reconstruct the building as one of the most important components of urban models is to identify and separate the building points from other points such as; terrain, trees, and vegetation. In addition, building roof segmentation is another important step in point cloud processing. In this paper, a multi-agent strategy is proposed for simultaneous extraction of building and segmentation of roof points from LiDAR point cloud. First, the ground and vegetation candidate points are removed from building points using local minimums of heights and returned pulse. Then different segments are extracted by analysis of the triangles formed on the remaining points by means of region growing method based on normal vectors. Finally, building segments are separated from other segments using area criterion and the united building segments are detected using a new method named ‘Grid Dilation’. The proposed method has been tested on the LiDAR data of the Vaihingen city, Germany. In addition to a visual interpretation, a quantitative assessment has been done. Due to lack of a ground truth, control points was selected manually from LiDAR point cloud and compared with points that classified using proposed method. The proposed method extracts the roof of buildings with an accuracy of 93%. Overall, the results indicate that proposed method could successfully identify the building segments without using additional resources, such as map or aerial photo. The main advantage of this method is its capability for extraction and segmentation of buildings containing parallel multi-level roof structures even with a very small height differences (e.g. 10 cm).

Urban planners and GIS users need to update their land use information in order to design and perform urban planning. Therefore the maps need to be updated regularly. An appropriate approach of updating them is using aerial images. Due to the high resolution of these images, their low cost, and also the possibility of aerial image acquisition in Iran, using aerial images for map updating has been a great concern. In order to update the available maps, the object changes should be understood, and then reconstructed. In this case and according to the properties of large scale maps (1/2000), buildings are very important features to be updated.In this research, a novel method based on Snake’s theory is proposed to detect buildings’ outlines. Also, a different Active Contour Model was used in order to compare the results of the proposed method. DSM and Ortho Images were produced using Stereo images and enough control points, the curve needed to initialize the snake model was obtained from the generated DSM. According to the results, the proposed method in this research provides the completeness of 86.043%, and the correctness of almost 91.124%.

Due to the growing cities، urban planning in the context of growing suburbanization، legal and illegal construction in urban is necessary. High-resolution stereo images that have been taken in two different times and the three-dimensional urban database، can help to building change detection. This research has been used to capability of stereo image to DSM generation and applying the difference digital surface model (D-DSM) in two periods، by using the image processing methods، Also according to existence of urban database and having geometric properties such as shape، size and height of urban features and applying the morphological filters، We attempted to D-DSM refinement and the elimination of non-Building features and the features extracted from this phase، as the building has been approved and will be reviewed to assess changes. With the evaluation results of D-DSM and set the suitable threshold، the buildings can be divided into three categories، unchanged، modified and the vague. After determining the vague things that can be done semi-automatically، the building change map provide and urban database be updated. In this study، we reached to 89. 0% completeness and 80. 8% correctness in semi-urban areas and 83. 9% completeness and 78. 2%correctness in urban areas. By examining the elevation of results was extracted، it can be divided the buildings to three categories: new Building، Reconstructed building and building with height changes (adding floor).

Building extraction becomes one of important fields of research in photogrammetry and machine vision. Involving airborne laser scanner (LiDAR) in geomatics, as an active sensor of 3D data acquisition, great evolution happened in data preparing to 3D outdoor object modeling. In consequence a great trend in developers and geomatics science researcher’s communities leaded them to develop techniques in extracting interesting objects from point cloud such as buildings. The LiDAR data don’t explicitly contain any geometric information and feature of objects. The feature such as planes, lines and corners can be only indirectly extracted by segmentation algorithms. In this research, a method for segmentation of LiDAR data for extracting buildings is proposed. In the proposed method, input data is an irregular LiDAR data and an adaptive clustering method for roof plane extraction from LiDAR data is implemented. So, to extract the roof structure, an assumption of planarity has been made. It is assumed that the roof can be modeled by a set of planar segments. In the proposed method the clustering is done without having any prior information about the number of the clusters. The clustering is done using FCM algorithm with considering maximum cluster number and then the extracted cluster will be modified using an iteratively split and merge technique. In this technique, small parallel planes are merged to create a new plane and a big plane created from different planes is divided to generate correct planes. This approach was evaluated using some buildings of used data set and the results prove high efficiency and reliability of this method for extraction of buildings.