Natural Neighbor Research Articles

A parameter-free hybrid instance selection algorithm based on local sets with natural neighbors

Instance selection aims to search for the best patterns in the training set and main instance selection methods include condensation methods, edition methods and hybrid methods. Hybrid methods combine advantages of both edition methods and condensation methods. Nevertheless, most of existing hybrid approaches heavily rely on parameters and are relatively time-consuming, resulting in the performance instability and application difficulty. Though several relatively fast and (or) parameter-free hybrid methods are proposed, they still have the difficulty in achieving both high accuracy and high reduction. In order to solve these problems, we present a new parameter-free hybrid instance selection algorithm based on local sets with natural neighbors (LSNaNIS). A new parameter-free definition for the local set is first proposed based on the fast search for natural neighbors. The new local set can fast and reasonably describe local characteristics of data. In LSNaNIS, we use the new local set to design an edition method (LSEdit) to remove harmful samples, a border method (LSBorder) to retain representative border samples and a core method (LSCore) to condense internal samples. Comparison experiments show that LSNaNIS is relatively fast and outperforms existing hybrid methods in improving the k-nearest neighbor in terms of both accuracy and reduction.

Can inpainting improve digital terrain analysis? Comparing techniques for void filling, surface reconstruction and geomorphometric analyses

AbstractThe investigation of form and processes in geomorphology and ecology is highly dependent on topographic data: a reliable digital terrain representation is in fact a key issue across environmental and earth sciences. In many cases, the processing of high‐resolution topographic data (e.g., light detection and ranging (LiDAR), structure from motion) has to face issues such as void filling, vegetation/feature removal and interpolation accuracy that are usually related to (i) intrinsic limitations of the adopted technology, (ii) local conditions affecting the survey or (iii) specific design scenario. In this paper, we develop a methodology to test the accuracy of an image inpainting algorithm to fill data voids in complex mountain areas. The devised experiment exploits the availability of a high‐resolution, LiDAR‐derived digital terrain model and the inpainting approach accuracy is checked against some widely used interpolation techniques (natural neighbor, spline, inverse distance weighting, kriging). In order to better mimic the actual surface texture, a methodology to introduce local topographic variability to the interpolated surface is also presented. The results show a better performance of the inpainting algorithm especially in the case of complex and rugged topography. Two examples showing an effective usage and accuracy of the proposed technique are reported, highlighting the drawbacks that a poor surface representation can introduce. The whole procedure is made freely available within a Matlab® script with the addition of sample files. ©2019 John Wiley & Sons, Ltd. ©2020 John Wiley & Sons, Ltd.

A novel graph-based clustering method using noise cutting

Recently, many methods have appeared in the field of cluster analysis. Most existing clustering algorithms have considerable limitations in dealing with local and nonlinear data patterns. Algorithms based on graphs provide good results for this problem. However, some widely used graph-based clustering methods, such as spectral clustering algorithms, are sensitive to noise and outliers. In this paper, a cut-point clustering algorithm (CutPC) based on a natural neighbor graph is proposed. The CutPC method performs noise cutting when a cut-point value is above the critical value. Normally, the method can automatically identify clusters with arbitrary shapes and detect outliers without any prior knowledge or preparatory parameter settings. The user can also adjust a coefficient to adapt clustering solutions for particular problems better. Experimental results on various synthetic and real-world datasets demonstrate the obvious superiority of CutPC compared with k-means, DBSCAN, DPC, SC, and DCore.

The effects of DEM interpolation on quantifying soil surface roughness using terrestrial LiDAR

Soil surface roughness (SSR) is often calculated based on the Digital Elevation Models (DEMs) obtained by interpolating points from terrestrial LiDAR measurements. This study aimed to investigate the effects of DEM interpolation and interpolation methods on quantification of SSR. A series of rainfall at the intensity of 59 and 178 mm hr−1 were simulated on a 2 by 6.1 m stony soil plot under 12 % and 20 % slope treatments. LiDAR measurements were conducted after each rainfall simulation at six positions around the soil plot. DEMs gridded at 5 and 10 mm were generated from LiDAR points using Inverse Distance Weighting, Natural Neighbors and Universal Kriging methods implemented in ESRI ArcGIS 10.5. Random roughness index (RR) was calculated based on the LiDAR-interpolated DEMs and LiDAR points directly and then compared. Results showed: 1) average values of RR from the DEMs were under-predicted compared to those directly calculated from LiDAR points by 7%–20%, indicating smoothing of the modeled surface during the interpolation process; 2) RR from LiDAR points indicated that soil surface became rougher as rainfall was progressed; 3) DEM errors increased as the surfaces evolved to rougher states and when 10 mm resolution was used; 4) the temporal SSR variations using the DEMs were -0.27 and -0.19 mm at 20 % slope and 0.30 and 0.36 mm at 12 % slope for the resolution 10 and 5 mm, respectively, while they were 0.22 mm at 20 % slope and 0.65 mm at 12 % slope when LiDAR points were directly used, the differences are due to the resulting increased DEM errors as the surface became rougher that masked the true changes in soil surface roughness. This study shows that at plot scale, DEMs generated through interpolations from LiDAR points underestimate soil surface roughness and are ineffective at tracking changes in soil surface roughness over time, and that LiDAR point data must be used instead at present scale.

Natural element method for axisymmetric elastoplasticproblems

Based on the meshless natural element method and incremental theory of plasticity, a new algorithm for elastoplastic analysis of a symmetric structure is put forward. The meshless method is a recently developed numerical method, which only requires the nodal information and can effectively overcome the difficulties brought about by the meshes. As widely known, conventional meshfree methods suffer from the numerical difficulties in dealing with the essential boundary conditions. Differing from most of meshfree methods, the natural element method (NEM) is a Galerkin-based meshless method that is built upon the notion of the natural neighbor interpolation. It is noteworthy that, there are two natural neighbor interpolants: Natural neighbor-based Sibson interpolation and Laplace interpolation (non-Sibsonian interpolation). The interpolation relies on the concepts of Delaunay triangulations and Dirichlet tessellations of a set of nodes to build the shape function. Compared with the moving least squares approximation widely used in many meshless methods, the natural neighbor interpolation does not involve complex matrix inversion, needs no artificial parameter but can improve the computational efficiency. Because the constructed shape functions from the natural neighbor interpolation possess the Kronecker delta function property, the essential boundary conditions can be imposed accurately without any special techniques. Meanwhile, the NEM does not require extra efforts to generate the background because it utilizes a set of Delaunay triangles, which are automatically identified in process of the basis function definition, as its background cell. Therefore, the NEM is a most promising numerical method which can be widely applied to nonlinear analysis of materials, three-dimensional calculation and large deformation problems. Generally, plasticity would make many projects more reasonable, and would fully utilize intensity potentially of structures. The NEM has been successfully used to solve elastoplastic problems of plane structures. In the present work, the NEM is applied to elastoplastic analysis of axisymmetric structures. Firstly, a brief description of the rudiments of Voronoi diagrams and Delaunay triangles in the context of natural neighbor interpolation is delineated. The basic principle of Sibson natural neighbor interpolation approach is also presented. A three-dimensional axisymmetric structure is created by the rotation of the cross section around the axis of symmetry and a set of properly scattered nodes within the cross section are employed. In the analysis of elastoplastic problems, the increments of stress and strain for axisymmetric problems are utilized to characterize the elastoplastic constitutive relationship. Owing to the nonlinear nature of plastic deformation, the computation at each load step employs an iterative solution method, either the Newton-Raphson method or the modified Newton-Raphson method. In order to reduce the expensive operations of stiffness matrix evaluation and factorization, the modified Newton-Raphson method with initial elastic stiffness matrix is adopted in the paper. The tangent predictor-radial return algorithm is applied to integrate the constitutive equations. The elastoplastic analyses of two classical axisymmetric structures have been successfully carried out and the fact that the present results show a good agreement with the analytical solutions demonstrate the effectiveness and accuracy of the proposed method for axisymmetric elastoplastic analyses.

Density peaks clustering algorithm based on natural nearest neighbours and its application in network advertising recognition

For the density peaks clustering algorithm, when dealing with multi-scale and manifold datasets, the algorithm cannot find the correct density peaks, and the distribution strategy is prone to the cascading effect of distribution errors. Hence, we propose a density peaks clustering algorithm based on natural nearest neighbours. In the proposed algorithm, the natural nearest neighbours of the sample are taken as the initial density by Gauss summation, and then it compares this initial density with the initial density of its k-nearest neighbours to form the final local density, the natural nearest neighbour information used to strengthen the relationship between samples, redefine the similarity measure between samples, and use it for remaining sample allocation. The experimental results on the synthetic datasets and network advertising recognition dataset show that the clustering effect of the proposed algorithm is better than that of DPC, DPCSA and FNDPCs, and more accurate advertising screening effect is achieved.

Density peaks clustering algorithm based on natural nearest neighbours and its application in network advertising recognition

For the density peaks clustering algorithm, when dealing with multi-scale and manifold datasets, the algorithm cannot find the correct density peaks, and the distribution strategy is prone to the cascading effect of distribution errors. Hence, we propose a density peaks clustering algorithm based on natural nearest neighbours. In the proposed algorithm, the natural nearest neighbours of the sample are taken as the initial density by Gauss summation, and then it compares this initial density with the initial density of its k-nearest neighbours to form the final local density, the natural nearest neighbour information used to strengthen the relationship between samples, redefine the similarity measure between samples, and use it for remaining sample allocation. The experimental results on the synthetic datasets and network advertising recognition dataset show that the clustering effect of the proposed algorithm is better than that of DPC, DPCSA and FNDPCs, and more accurate advertising screening effect is achieved.

ELS: A Fast Parameter-Free Edition Algorithm With Natural Neighbors-Based Local Sets for k Nearest Neighbor

The instance reduction is one of the data preprocessing methods and aims to remove noises and (or) redundant instances from the training set. In the instance reduction, one of the most representative techniques is the edition method which can remove harmful instances from the training set to improve the prediction accuracy of k nearest neighbor (KNN). Nevertheless, most of existing edition methods still have some drawbacks, such as the parameter dependency, high computational time and relatively low accuracy. To solve these problems, we present a new fast parameter-free edition method based on local sets with natural neighbors (ELS). In ELS, we define a new concept of local sets by introducing natural neighbors. ELS can use the local sets to keep more reasonable class boundaries and effectively filtering out noisy instances (including global outliers). The main advantages of ELS are that (a) it is parameter-free; (b) it can remove global outliers and noisy instances; (c) it is relatively fast. Experiments clearly verify that (a) ELS outperforms existing representative edition methods in improving the prediction accuracy of KNN; (b) ELS can improve the performance of the condensation method and hybrid method in terms of both accuracy and reduction; (c) ELS consumes relatively low running time.

Universal Kriging Prediction of Line-of-Sight Microwave Fading

Prediction of the severity of multipath fading is fundamental to the design of point-to-point terrestrial fixed microwave links at frequencies below 10 GHz, but error in this prediction may be significant in countries such as Australia, not represented in the dataset used to generate existing empirical models. We take advantage of recently collected worst-month fading data from Australia, and find new parameters particularly useful in predicting the severe fading experienced in Northern Australia. These parameters are from very irregularly spaced weather stations, so we investigate various interpolation techniques for this situation, including a new version of natural neighbour interpolation. Conventional multipath prediction models are based on ordinary least squares (OLS) regression, but we refine this, taking spatial correlation into account with generalised least squares (GLS) regression. We then demonstrate further improvement in regions well populated by measured data, by employing universal kriging.

Satellite-Based Fire Progression Mapping: A Comprehensive Assessment for Large Fires in Northern California

Satellite-based active fire (AF) products provide opportunities for constructing continuous fire progression maps, a critical dataset needed for improved fire behavior modeling and fire management. This study aims to investigate the geospatial interpolation techniques in mapping the daily fire progression and assess the accuracy of the derived maps from multisensor AF products. We focused on 42 large wildfires greater than 5000 acres in Northern California from 2017 to 2018, where the USDA Forest Service National Infrared Operations (NIROPS) daily fire perimeters were available for the comparison. The standard AF products from the moderate resolution imaging spectroradiometer (MODIS), the visible infrared imaging radiometer suite (VIIRS), and the combined products were used as inputs. We found that the estimated fire progression areas generated by the natural neighbor method with the combined MODIS and VIIRS AF input layers performed the best, with R 2 of 0.7 ± 0.31 and RMSE of 1.25 ± 1.21 (103 acres) at a daily time scale; the accuracy was higher when assessed at a two-day rolling window, e.g., R 2 of 0.83 ± 0.20 and RMSE of 0.74 ± 0.94 (103 acres). A relatively higher spatial accuracy was found using the 375 m VIIRS AF product as inputs, with a kappa score of 0.55 and an overall accuracy score of 0.59, when interpolated with the natural neighbor method. Furthermore, the locational pixel-based comparison showed 61% matched to a single day and an additional 25% explained within ±1 day of the estimation, revealing greater confidence in fire progression estimation at a two-day moving time interval. This study demonstrated the efficacy and potential improvements of daily fire progression mapping at local and regional scales.

Analyzing single-lap joints bonded with a brittle adhesive by an elastic meshless method

Adhesive bonding is a widespread joining technique in the industry, namely aeronautical and automotive. However, the associated products often have complex and non-standard shapes. Numerical modeling, such as that based on the finite element method (FEM), is commonly applied for bonded joint design, although there are still limitations linked to the method. Meshless methods (MM) are a viable option to overcome some limitations of the FEM. However, these methods are still in development.In this work, a robust meshless method, the Natural Neighbor Radial Point Interpolation Method (NNRPIM), was applied to the analysis of adhesive joints. First, experimental data of single-lap joints (SLJ) corresponding to four overlap lengths (LO) and a brittle adhesive were measured as a benchmark. Then, all the joint geometries were simulated utilizing the FEM and NNRPIM methodologies. The NNRPIM proved to be a good alternative to the FEM, providing similar strength prediction with differences less than 3% between them. Moreover, the stress distribution curves were compared, and differences of 6% in the peak shear stresses were found. In conclusion, the NNRPIM provides accurate results and could be utilized for further study of adhesively bonded joints.

Considerações sobre o LiDAR na análise de instabilidade de vertentes

The digital models of landform representation allow several types of geomorphological analysis, such as assessing prone areas to the mass movement. Among the data set that applied to elaborate such models, LiDAR (Light Detection and Ranging) technology has stood out due to its capacity to represent in detail terrain specific characteristics. This work aimed to present a variety of studies based on LiDAR to evaluate slope instability and to describe an interpolation test of LiDAR raw data to generate a Digital Terrain Model considering different interpolation approaches, in an area affected by shallow landslides. The results show that the tool applied is efficient when combined Triangulation to Natural Neighbor methods.

Spectral Clustering Algorithm Based on Fast Search of Natural Neighbors

The spectral clustering is a typical and efficient clustering algorithm. However, the performance of spectral algorithm depends on the determination of the appropriate similarity matrix and the number of clusters. We propose a new spectral clustering algorithm based on fast search of natural neighbors (FSNN-SC) in this paper. In the algorithm, we design a fast search algorithm to obtain the natural characteristic value supk of natural neighbor algorithm in order to improve the efficiency of searching neighbors and to construct a high-quality similarity matrix. At the same time, we design a deep traversal algorithm to adaptively determine the cluster number C. The experimental results verify that our methods are able to improve the search efficiency and find correct number of clusters. The compared experiments show that the accuracy and efficiency of the proposed algorithm are better than others.

A B-Spline Method With AIS Optimization for 2-D IoT-Based Overpressure Reconstruction

In blast wave monitoring, a traditional travel time tomography method is encountered with local minimum travel time and low coverage density of rays. In this article, a novel B-spline fitting method with the knot-optimization artificial immune system (AIS) is proposed for 2-D overpressure reconstruction. It possesses the advantages of handling point sets of large sizes and adjusts the knot vector flexibly. Based on the overpressure value in the explosion from the travel time tomography method, the proposed method combining the advantages of B-splines and knot point optimization AIS is able to achieve the optimal sensor distribution and raise the reconstruction precision. The detailed experimental results about the comparison of linear fitting interpolation, cubic fitting interpolation, natural neighbor fitting interpolation, v4 fitting interpolation, Delaunay triangulation fitting, and B-spline method are also given. Furthermore, for the knot optimization issue in B-spline, the proposed adaptive fitting method with knot-optimization AIS has a smaller root-mean-square (RMS) error with eight knot nodes in comparison with the classic B-spline fitting method. This article is conducted to provide new insights to reconstructing 2-D Internet-of-Things-based (IoT-based) overpressure in blast wave monitoring more precisely under limited sensor deployment and further give a new approach to overpressure reconstruction scenarios.

Comparison of Different DEM Generation Methods based on Open Source Datasets

Digital Elevation Model (DEM) is one of the developed techniques for relief representation. The definition of a DEM construction is the modeling technique of earth surface from existing data. DEM plays a role as one of the fundamental information requirement that has been generally utilized in GIS data structures. The main aim of this research is to present a methodology for assessing DEMs generation methods. The DEMs data will be extracted from open source data e.g. Google Earth. The tested data will be compared with data produced from formal institutions such as General Directorate of Surveying. The study area has been chosen in south of Iraq (Al-Gharraf / Dhi Qar governorate. The methods of DEMs creation are kriging, IDW (inverse distance weight), spline, and natural neighbor. This research used different software for processing and analysis such as ArcGIS 10.2, TCX and Civil 3D. Two- sample t-test has been adopted to investigate the mean of elevation differences between compared datasets. The results showed that the spline is the best method that can be used to build DEM in this study area.

A feature-preserving point cloud denoising algorithm for LiDAR-derived DEM construction

To attenuate positional errors of LiDAR-derived datasets for constructing digital elevation models (DEMs), a feature-preserving point denoising algorithm (F-PDA) is developed in this paper. F-PDA includes three main steps: surface normal estimation, normal filtering and point position update. Numerical tests with two simulated surfaces indicate that F-PDA is always more accurate than kriging and natural neighbour. Furthermore, F-PDA has a high effectiveness of preserving feature lines. Real-world examples of interpolating LiDAR samples demonstrate that F-PDA can best retain both prominent and subtle terrain features, while faithfully removing errors in mountainous and flat regions. Moreover, it outperforms some well-known interpolation methods.

Suitability Calculation for Red Spicy Pepper Cultivation (Capsicum annum L.) Using Hybrid GIS-Based Multicriteria Analysis

Red spicy pepper is traditionally considered as the fundamental ingredient for multiple authentic products of Eastern Croatia. The objectives of this study were to: (1) evaluate the optimal interpolation method necessary for modeling of criteria layers; (2) calculate the sustainability and vulnerability of red spicy pepper cultivation using hybrid Geographic Information System (GIS)-based multicriteria analysis with the analytical hierarchy process (AHP) method; (3) determine the suitability classes for red spicy pepper cultivation using K-means unsupervised classification. The inverse distance weighted interpolation method was selected as optimal as it produced higher accuracies than ordinary kriging and natural neighbour. Sustainability and vulnerability represented the positive and negative influences on red spicy pepper production. These values served as the input in the K-means unsupervised classification of four classes. Classes were ranked by the average of mean class sustainability and vulnerability values. Top two ranked classes, highest suitability and moderate-high suitability, produced suitability values of 3.618 and 3.477 out of a possible 4.000, respectively. These classes were considered as the most suitable for red spicy pepper cultivation, covering an area of 2167.5 ha (6.9% of the total study area). A suitability map for red spicy pepper cultivation was created as a basis for the establishment of red spicy pepper plantations.

Research and Experimental Comparison of Topographic Modeling Methods

The topography of the earth's surface can be represented in GIS by DEM data. Surface modeling is the process of determining natural or artificial surfaces using one or more mathematical equations. A general surface modeling algorithm for all applications is not available, each method of creating a topographic surface has several advantages, disadvantages and depends on its processing direction. As such, experimenting, evaluating and selecting algorithms that are appropriate to the reality of the data and research area are necessary. Research paper, experimentally evaluating 4 Spline, IDW, Kriging and Natural Neighbor algorithms to model terrain on two map pieces representing different terrain types, the research results show that: the adapting each algorithm depends heavily on the terrain of each interpolation area. Spline interpolated terrain surfaces in more detail in ravine and valley areas; Natural Neighbor excels at matching the contours of data in all regions. IDW and Kriging algorithms have similar results and have lower accuracy than the above two methods, especially it is difficult to interpolate peaks and slopes. MAE, respectively, of high and medium hills and mountains are Spline (9.7, 10.3), NN (11.8, 10.1), IDW (13.0, 10.9), Kriging (13.3, 12.2).
 Spatial interpolation, topographic modeling, DEM, DSM, accuracy

Latest Geodetic Changes of Austre Lovénbreen and Pedersenbreen, Svalbard

Geodetic mass changes in the Svalbard glaciers Austre Lovénbreen and Pedersenbreen were studied via high-precision real-time kinematic (RTK)-global positioning system (GPS) measurements from 2013 to 2015. To evaluate the elevation changes of the two Svalbard glaciers, more than 10,000 GPS records for each glacier surface were collected every year from 2013 to 2015. The results of several widely used interpolation methods (i.e., inverse distance weighting (IDW), ordinary kriging (OK), universal kriging (UK), natural neighbor (NN), spline interpolation, and Topo to Raster (TTR) interpolation) were compared. Considering the smoothness and accuracy of the glacier surface, NN interpolation was selected as the most suitable interpolation method to generate a surface digital elevation model (DEM). In addition, we compared two procedures for calculating elevation changes: using DEMs generated from the direct interpolation of the RTK-GPS points and using the elevation bias of crossover points from the RTK-GPS tracks in different years. Then, the geodetic mass balances were calculated by converting the elevation changes to their water equivalents. Comparing the geodetic mass balances calculated with and without considering snow depth revealed that ignoring the effect of snow depth, which differs greatly over a short time interval, might lead to bias in mass balance investigation. In summary, there was a positive correlation between the geodetic mass balance and the corresponding elevation. The mass loss increased with decreasing elevation, and the mean annual gradients of the geodetic mass balance along the elevation of Austre Lovénbreen and Pedersenbreen in 2013–2015 were approximately 2.60‰ and 2.35‰, respectively. The gradients at the glacier snouts were three times larger than those over the whole glaciers. Additionally, some mass gain occurred in certain high-elevation regions. Compared with a 2019 DEM generated from unmanned aerial vehicle measurement, the glacier snout areas presented an accelerating thinning situation in 2015–2019.

HUBUNGAN SPASIAL TRANSAKSI TANAH DENGAN PERKEMBANGAN WILAYAH DI KABUPATEN CIREBON, JAWA BARAT

The big potential of Cirebon Regency with good transportation access and demography bonus is able to promote its regional development. Not all of the potential has been optimized to invite investment in Cirebon Regency. One of the causes is some of the lands haven't been registered, so that obstructs investment that requires land. We explore the spatial relationship between the regional development level and land transaction pattern. We calculate the entropy diversification index to see the level of regional development. We also interpolate the land transaction to see the spatial pattern. The natural neighbor interpolation method was used to show spatial patterns of land transactions. Then, we overlay the map of the entropy index and land transaction pattern map to see the relationship spatially. Land transactions in Cirebon Regency gathered in a very developed and developing region. Districts of Palimanan, Depok, Plumbon, Weru, and Plered that are categorized very developed have more quantity of transactions and land transaction value than the undeveloped area. Pasaleman district has Kecamatan Depok, has the smallest entropy index, and categorized as undeveloped has a lower number and value of land transactions. We found a similar spatial pattern between the regional level of development and land transaction interpolation.