Using range data in automatic modeling of buildings

Planar surface patches, extracted for instance from a captured 3D point cloud, play an important role as a feature in modeling and/or recognizing objects and environments. In terms of extracting such planar surface patches, the main technical issue involved may be of overcoming the trade-off between the modeling accuracy and the computational efficiency, especially, when the 3D point cloud captured is noisy. Conventionally, they have taken approaches that seek either for high computational efficiency at the expense of modeling accuracy or vice versa. This paper contributes the advancement in the methodology of extracting planar surface patches from a noisy 3D point cloud by presenting a method that provides high modeling accuracy yet under low computational cost. The major contribution of the proposed method is summarized as follows: 1) A robust estimation of surface normal vectors in such a way as to minimize the effect of noise in the data. 2) An accurate localization of density peaks based on a spherical coordinate operator with a flexible size of sliding window used for estimating the density of surface normal vectors represented on a unit spherical surface as a means of accurately identifying those planar surface patches of the same orientation. 3) A segmentation of individual planar surface patches of the same orientation by projecting those 3D points clustered as belonging to a same peak onto x, y and z axes of the Cartesian coordinate with the Y axis representing their orientation. The experimental results show the effectiveness of the proposed method in comparison with conventional ones for various indoor and outdoor images.

Background. Modern methods for determining the coordinates of elements of wireless sensor networks allow solving the problems of determining the mutual distances between the elements of a wireless network under the assumption that the errors in measuring the mutual distances between network elements are distributed according to the normal law. With increasing requirements for the accuracy of determining coordinates, these methods do not allow solving the problem.
 Objective. The purpose of the paper is to improve the accuracy of determining the coordinates of elements of wireless sensor networks by using robust estimation methods.
 Methods. Determination of coordinates of elements of wireless sensor networks is implemented on the basis of two robust methods.
 The first is the use of a median estimate based on multiple measurements of the mutual distances between elements to determine their coordinates.
 The second is based on multiple measurements of the mutual distances between elements, determining their coordinates based on the Huber influence function and comparing two robust methods.
 Results. The use of a robust method based on the Huber influence function makes it possible to increase the accuracy of determining the coordinates of elements of wireless sensor networks by 5-10% compared to classical estimation methods.
 Conclusions. The proposed robust approach to determining the coordinates of elements of wireless sensor networks can be implemented in modern ground-based sensor networks for various purposes.
 Keywords: wireless sensor network; determination of coordinates of elements of wireless sensor networks; robust estimation methods.

Localization of an unmanned ground vehicle (UGV) is a very important task for autonomous vehicle navigation. In this paper, we propose a computer vision technique to identify the location of an outdoor UGV. The proposed technique is based on 3D registration of 360 degree laser range data to a digital surface model (DSM). A long sequence of range frames is obtained from a rotating range sensor which is mounted on the top of the vehicle. Two novel approaches are proposed for accurate 3D registration of range data and the DSM. First, registration is done between range frames in a pair-wise manner followed by a refinement with the DSM. Second, we divide the DSM to several layers and find correspondences near the current vehicle elevation. This reduces the number of outliers and facilitates fast localization. Experimental results show that the proposed approaches yield better performance in 3D localization compared to conventional 3D registration techniques. Error analysis on four outdoor paths is presented with respect to ground truth.

Automated Crater detection from Co-registered optical images, elevation maps and slope maps using deep learning

The linear regression model is one of the most important basic statistical models widely used in data analysis in many fields, such as economics, social sciences, and medicine. This model used to illustrate the relationship between a dependent variable and several independent variables. Although the ordinary least squares (OLS) method is effective for estimation under appropriate conditions, the presence of outliers or extreme values ​​may reduce the effectiveness and accuracy of these estimates. Therefore, the need has emerged to use alternative and more robust estimation methods to address these observations, known as robust estimation methods. This study compares the classical ordinary least squares method with several robust methods, such as the S-estimation, M-estimation, MM-estimation, and Huber methods, to evaluate their performance in the presence of data contamination. The analysis results showed that the Huber method outperformed the other methods when contamination levels reached 7% or more, making it an effective choice for obtaining more accurate and reliable estimates in environments containing imperfect or contaminated data.

Robust estimation methods can efficiently eliminate or weaken the effects of gross errors on parameter estimation when such errors exist in observations. Remainder reliability can essentially show whether gross errors may be eliminated or weakened by robust estimation. Taking 8 leveling networks and 4 trilateration networks with different remainder reliability adjustments as examples, simulation experiments are used to compare 4 frequently used robust estimation methods. The results indicate that when observations simultaneously contain two gross errors and if the remainder reliability RR20.45, all the four robust estimation methods may effectively eliminate or weaken these effects.

Three-dimensional (3D) mapping deals with the problem of building a map of the unknown environments explored by a mobile robot. In contrast to 2D maps, 3D maps contain richer information of the visited places. Besides enabling robot navigation in 3D, a 3D map of the robot surroundings could be of great importance for higher-level robotic tasks, like scene interpretation and object interaction or manipulation, as well as for visualization purposes in general, which are required in surveillance, urban search and rescue, surveying, and others. Hence, the goal of this thesis is to develop a system which is capable of reconstructing the surrounding environment of a mobile robot as a three-dimensional map. Microsoft Kinect camera is a novel sensing sensor that captures dense depth images along with RGB images at high frame rate. Recently, it has dominated the stage of 3D robotic sensing, as it is low-cost, low-power. For this work, it is used as the exteroceptive sensor and obtains 3D point clouds of the surrounding environment. Meanwhile, the wheel odometry of the robot is used to initialize the search for correspondences between different observations. As a single 3D point cloud generated by the Microsoft Kinect sensor is composed of many tens of thousands of data points, it is necessary to compress the raw data to process them efficiently. The method chosen in this work is to use a feature-based representation which simplifies the 3D mapping procedure. The chosen features are planar surfaces and orthogonal corners, which is based on the fact that indoor environments are designed such that walls, ground floors, pillars, and other major parts of the building structures can be modeled as planar surface patches, which are parallel or perpendicular to each other. While orthogonal corners are presented as higher features which are more distinguishable in indoor environment. In this thesis, the main idea is to obtain spatial constraints between pairwise frames by building correspondences between the extracted vertical plane features and corner features. A plane matching algorithm is presented that maximizes the similarity metric between a pair of planes within a search space to determine correspondences between planes. The corner matching result is based on the plane matching results. The estimated spatial constraints form the edges of a pose graph, referred to as graph-based SLAM front-end. In order to build a map, however, a robot must be able to recognize places that it has previously visited. Limitations in sensor processing problem, coupled with environmental ambiguity, make this difficult. In this thesis, we describe a loop closure detection algorithm by compressing point clouds into viewpoint feature histograms, inspired by their strong recognition ability. The estimated roto-translation between detected loop frames is added to the graph representing this newly discovered constraint. Due to the estimation errors, the estimated edges form a non-globally consistent trajectory. With the aid of a linear pose graph optimizing algorithm, the most likely configuration of the robot poses can be estimated given the edges of the graph, referred to as SLAM back-end. Finally, the 3D map is retrieved by attaching each acquired point cloud to the corresponding pose estimate. The approach is validated through different experiments with a mobile robot in an indoor environment

Canonical correlation analysis is one of the common methods for analyzing data and know the relationship between two sets of variables under study, as it depends on the process of analyzing the variance matrix or the correlation matrix. Researchers resort to the use of many methods to estimate canonical correlation (CC); some are biased for outliers, and others are resistant to those values; in addition, there are standards that check the efficiency of estimation methods.
 In our research, we dealt with robust estimation methods that depend on the correlation matrix in the analysis process to obtain a robust canonical correlation coefficient, which is the method of Biweight Midcorrelation coefficient (Bi) and Kendall-tau correlation coefficient (Ke).
 From the comparison between these two methods through the empirical influence function with standard scaled and transformed estimator, the results indicated the efficiency and the preference of the (Bi) method. The study also has application with real data followed a multivariate normal distribution with two sets; the first group represents monthly averages for quantities of exported oil from three OPEC countries, namely Saudi Arabia, Iraq, and Kuwait, the other group represents the returns of those quantities for the period from 2015 to 2019, after applied (Bi) method and estimate IF, the strongest influence about CC was at thirty four-months and the lowest was at twenty-seven

This paper deals with the automated creation of geometric and photometric correct 3-D models of the world. Those models can be used for virtual reality, tele-presence, digital cinematography and urban planning applications. The combination of range (dense depth estimates) and image sensing (color information) provides data-sets which allow us to create geometrically correct, photorealistic models of high quality. The 3-D models are first built from range data using a volumetric set intersection method previously developed by us. Photometry can be napped onto these models by registering features from both the 3-D and 2-D data sets. Range data segmentation algorithms have been developed to identify planar regions, determine linear features from planar intersections that can serve as features for registration with 2-D imagery lines, and reduce the overall complexity of the models. Results are shown for building models of large buildings on our campus using real data acquired from multiple sensors.

This paper presents an algorithm for extracting planar patches by integrating both intensity and range data provided by a stereo system. For dealing with noisy and sparse range data, the initial segmentation is based on intensity information, and then the resulted regions are thresholded using depth data. This new algorithm, different from the existing ones that use only range data in the segmentation process, produces accurate planar patches that are then used for building a panoramic image-based model for mobile robot navigation.

Automatic 3D building reconstruction from multi-view aerial images with deep learning

In order to construct the city three dimensional (3D) building model, building contours must be extracted in fast and accurately. Traditional engineering survey and photogrammetry method is low efficient to do this. The Light Detect and Ranging (LiDAR), so called Airborne Laser Scanning, works on the same principle as radar except using light waves instead of radio waves. It can measure distance and get high density raw laser point data (so called point cloud) by the way of scanning ground objects. We can produce city Digital Surface Model (DSM) in a fast rapid via LiDAR point cloud and modern computer. And the DSM is an important basis to build 3D digital city. Due to the LiDAR point cloud data are noncontinuity in distribution, interval and in density, it is difficult to extracted buildings contours from the single point cloud data. In this paper, we extract urban building contours to use digital image processing technology from the airborne LiDAR points cloud data without any assistance data. This way can rapidly extract building contours with low cost in a large urban area. The result shows it is universally applicable for extract urban building contours from raw LiDAR point cloud.

Effectiveness of chosen robust estimation methods compared to the level of network reliability The work presents the results of studies on dependence of effectiveness of chosen robust estimation methods from the internal reliability level of a geodetic network. The studies use computer-simulated observation systems, so it was possible to analyse many variants differing from each other in a planned way. Four methods of robust estimation have been chosen for the studies, differing substantially in the approach to weight modifications. For comparative reasons, the effectiveness studies have also been conducted for the very popular method in surveying practice, of gross error detection basing on LS estimation results, the so called iterative data snooping. The studies show that there is a relation between the level of network internal reliability and the effectiveness of robust estimation methods. In most cases, in which the observation contaminated by a gross error was characterized by a low index of internal reliability, the robust estimation led to results being essentially far from expectations.

A scale-adaptive digital elevation model (S-DEM) method is proposed for multi-scale terrain analysis using a single high-resolution digital elevation model (DEM) database. The motivation is to construct a DEM that is self-adaptive to a given scale of an application, rather than letting the application fit into the built-in scale of the DEM. The method is based on an adaptive compound point extraction (CPE) algorithm that extracts surface ‘significant points’ from a high-resolution DEM according to their degree of importance (DOI) to the scale of an application. A data structure can be established to match the demand from an application at a coarser scale. Based on the data structure, a triangulated irregular network (TIN) model can be generated to support the terrain analysis at the desired scale. The aim of the S-DEM is to support multi-scale applications in three aspects, namely, ‘one database for all scales and scale-adaptive’ (i.e. matching any application scale using a single high-resolution DEM), ‘consistent measurement’ (i.e. delivering more constant measurements of terrain parameters with changing scales), and ‘skeleton preservation’ (i.e. preserving basic streamlines with changing scales). Compared with the raster resampling algorithm and the maximum z-tolerance algorithm, we find that the proposed method offers better performance, providing values that meet the accuracy requirements set by DEM data standards for different scales, and producing analytical derivatives that retain terrain features with consistent measurements of terrain parameters.

An Arctic underwater terrain matching method integrating template matching and DEM super-resolution