Figure - available from: Earth Science Informatics
This content is subject to copyright. Terms and conditions apply.
Source publication
There are substantial problems in the photogrammetric image matching especially in the images taken by UAVs in the regions where grassland, waterfront, forest, buildings, bridges, high-voltage lines etc. in the urban and rural areas. The main reason of these problems are color, tone, texture, contrast and scale changes cannot be successfully detect...
Citations
... The Thin Plate Spline (TPS) is a specific spline used in interpolation and surface adjustment (Duchon, 1977). TPS is used in problems where surface smoothness is required, such as deformation mapping (Donato and Belongie, 2002), three-dimensional reconstruction (Sokolov et al., 2017), and other image processing applications (Atik et al., 2020). ...
With the advancement of artificial intelligence and embedded hardware development, the utilization of various autonomous navigation methods for mobile robots has become increasingly feasible. Consequently, the need for robust validation methodologies for these locomotion methods has arisen. This paper presents a novel ground truth positioning collection method relying on computer vision. In this method, a camera is positioned overhead to detect the robot's position through a computer vision technique. The image used to retrieve the positioning ground truth is collected synchronously with data from other sensors. By considering the camera-derived position as the ground truth, a comparative analysis can be conducted to develop, analyze, and test different robot odometry methods. In addition to proposing the ground truth collection methodology in this article, we also compare using a DNN to perform odometry using data from different sensors as input. The results demonstrate the efficacy of our ground truth collection method in assessing and comparing different odometry methods for mobile robots. This research contributes to the field of mobile robotics by offering a reliable and versatile approach to assess and compare odometry techniques, which is crucial for developing and deploying autonomous robotic systems.
... Thin plate splines (TPS) is usually a radial basis function (RBF) that interpolates using distances. Such functions are mostly used for multivariate (n-dimensional) data approximation (Buhmann, 2003) The main approach of the TPS is to calculate interpolation parameters using GCPs for the coordinate transformation of other points (Atik et al., 2020). TPS are robust against local distortions. ...
Remote sensing enables the measurement, extraction and presentation of useful information at various spatial and temporal scales. It is used by decision-makers to create sustainable projects. However, the high geometric accuracy of satellite images is vital for the accurate planning of sustainable projects and for accurately extracting information from remote sensing data. The geometric correction process for obtaining orthoimages requires a digital elevation model (DEM), ground control points (GCP) common in the object and image space, and a model that represents the transformation between the object space and the image space. Therefore, the accuracy of an orthoimage depends on the distribution of the ground control points, the model used, and the precision of the digital elevation model. In this study, the effect of the number and distribution of ground control points on the accuracy of the polynomial transformation model, rational function model and thin plate spline methods used in obtaining the orthoimage was investigated. The performance of the methods was evaluated by using a very high-resolution Pléiades-1B satellite image. The digital elevation model (DEM) was obtained by the photogrammetric method using aerial photographs. Experimental results demonstrate that the appropriate GCP distribution significantly improved the geometric correction accuracy of the orthoimages.
... To overcome this problem, powerful outlier removal techniques are designed that are capable of generating putative keypoints correspondences. Amongst these, RANSAC (Fischler and Bolles 1981) is considered to be the classical statistical parameter estimation technique, which removes the wrongly matched points, iteratively generates optimal subset of matched keypoints, and computes transformation parameters for image co-registration (Misra et al. 2012;Atik et al. 2020). In literature, multiple variants of RANSAC algorithm are presented, which are claimed to be fast, efficient and sometimes more effective for particular types of images. ...
Remote sensing image mosaicking is an essential processing step in generating large area coverage map using multi-temporal image scenes/strips. The mosaic data is useful for various space-borne applications that span across national level crop assessment, wetland monitoring, and snow and glacier studies, to derive important environmental indicators for sustainable development. This article highlights a novel image mosaicking processing workflow that ingests input geo-referenced image strips with sufficient overlap in-between, and generates country-level mosaic data product. The procedure takes care of large-sized geo-referenced image’s handling and re-projection, and makes data ready for mosaic processing. We have developed strip geo-registration method using Scale Invariant Feature Transform (SIFT) and Mode Biased Random Sample Consensus (MB-RANSAC) outlier removal technique to achieve sub-pixel registration accuracy. Image stitching workflow ingests co-registered image strips, and performs overlap extraction, seamline detection using multi-frame joint strategy, and image blending using region-based statistics in an automatic manner. The mosaic system has been evaluated with Resourcesat’s medium resolution optical remote sensing images over Indian subcontinent, and it has been confirmed that the common region among image strips attains required radiometric and geometric fidelity after correction. It also has been found that the average spectra deviation is less than 0.127% at different classes.
... The initial estimated values are then iteratively improved using non-linear least squares adjustment [18]. For this, at least two photographs of the same object are used [19]. In this study, DJI Phantom 4 Pro UAV platform is used to capture the high-resolution images in Istanbul Technical University Ayazaga Campus. ...
... The initial estimated values are then iteratively improved using non-linear least squares adjustment [18]. For this, at least two photographs of the same object are used [19]. In terms of alignment photographs, 3.367 aerial photos were aligned using automatically detected features from each image via the software. ...
... Metric values obtained in machine learning are calculated with a confusion matrix. The equations of the metrics are shown in Equations (16)- (19). ...
With the development of photogrammetry technologies, point clouds have found a wide range of use in academic and commercial areas. This situation has made it essential to extract information from point clouds. In particular, artificial intelligence applications have been used to extract information from point clouds to complex structures. Point cloud classification is also one of the leading areas where these applications are used. In this study, the classification of point clouds obtained by aerial photogrammetry and Light Detection and Ranging (LiDAR) technology belonging to the same region is performed by using machine learning. For this purpose, nine popular machine learning methods have been used. Geometric features obtained from point clouds were used for the feature spaces created for classification. Color information is also added to these in the photogrammetric point cloud. According to the LiDAR point cloud results, the highest overall accuracies were obtained as 0.96 with the Multilayer Perceptron (MLP) method. The lowest overall accuracies were obtained as 0.50 with the AdaBoost method. The method with the highest overall accuracy was achieved with the MLP (0.90) method. The lowest overall accuracy method is the GNB method with 0.25 overall accuracy.
Digital Elevation Models (DEM) obtained by different sources are used in many applications that require height information. With the recent development of LIDAR (Light Detection and Ranging) technology, the aerial LIDAR technique is commonly used to collect point cloud data for producing DEM in cm accuracy. In the study, DEMs were created with five different interpolation methods using the point clouds. Firstly, DEM was produced with point groups of two different densities selected from Google Earth Pro. Point clouds obtained from airborne LIDAR were used as reference data. Thus, the effect of different point densities and interpolation methods on DEM production with Google Earth Pro was examined. Additionally, DEMs obtained with LIDAR were compared with SRTM (Shuttle Radar Topography Mission) and ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer) DEMs. DEMs obtained from Google Earth PRO give better results than the other two DEMs. DEMs produced with GEP achieve an accuracy of 3.44 m, while the accuracy of SRTM and ASTER GDEM data is 4.05 m and 5.88 m, respectively.
