PRISM :: Browsing by Author "Wang, Ruisheng"

Browsing by Author "Wang, Ruisheng"

Now showing 1 - 20 of 29

Open Access
2D-3D Registration for a High-speed Biplanar Videoradiography Imaging System
(2022-04) Zhang, Shu; Lichti, Derek; Detchev, Ivan; Ronsky, Janet; Wang, Ruisheng; Armenakis, Costas; Lichti, Derek
High-Speed Biplanar Videoradiography (HSBV) is an X-ray based imaging system that can derive dynamic bony translations and rotations. The 2D-3D registration process matches a 3D bone model acquired from magnetic resonance imaging (MRI) or computed tomography (CT) scans with the 2D X-ray image pairs. 2D-3D registration is usually conducted in two ways, marker-based and model-based registration. The marker-based method is known for its high registration accuracy thanks to corresponding marker pairs. On the other hand, the model-based method avoids the implantation of radiopaque beads but uses the radiograph’s features, intensities, or gradients to accomplish the data alignment. Two novel marker-based registration methods, the back-projection and the projection methods, were proposed and compared with the state-of-the-art RSA (Roentgen Stereophotogrammetric Analysis) method. A 3D printed bone model with beads was used to validate the proposed methods. The results showed that both methods acquired higher accuracy than the RSA method. In addition, the projection and back-projection techniques can be used for the model-based registration while the RSA method cannot. The projection method was applied to a model-based registration to achieve higher accuracy, providing a 3D reconstruction accuracy of 0.79 mm for both the tibia and femur. By using the non-rigid transformation with a scale factor, this accuracy was successfully increased to 0.56 mm for the tibia and 0.64 mm for the femur. The discrepancies in the 2D-3D registration that led to the non-rigid transformation were validated. It was caused by the offset between the detected edge points in the radiographs and their actual position. A Kalman filter was tested on the marker- and model-based registration results with different random processes and parameters. For marker-based registrations, the standard deviations of the kinematics parameters were improved by 25 – 62% for the translations and 35 – 43% for the rotations. For the model-based registration, these standard deviations were improved by 6 – 38% and 29 – 38%, respectively. While the projection method provided higher accuracy, the back-projection method had the larger capture range for the initialization. An automatic initialization method with 64 starting poses based on the back-projection method was proposed and validated. It successfully eliminated the user intervention in the registration initialization. The improved 2D-3D registration with non-rigid transformation and dynamic estimation allows the determination of accurate 3D kinematic parameters with high efficiency. These kinematic parameters can be used to calculate joint cartilage contact mechanics that provide insight into the mechanical processes and mechanisms of joint degeneration or pathology.
Open Access
3D Reconstruction of Building Interiors Using Point Clouds
(2018-04-24) Xie, Lei; Wang, Ruisheng; Shahbazi, Mozhdeh; Hassan, Quazi
The automatic modeling of as-built building interiors, known as indoor building reconstruction, is gaining increasing attention because of its widespread applications. With the development of sensors to acquire high-quality point clouds, a new modeling scheme called scan-to-BIM (building information modeling) emerged. However, the traditional scan-to-BIM process is time tedious and labor intensive. Most existing automatic indoor building reconstruction solutions can only fit the specific data or lack of detailed model representation. In this thesis, we propose two automatic reconstruction methods from 2D linear primitives and 3D planar primitives respectively, to create 2D floor plans and 3D building models. The approach using 2D primitives is well suited for large-scale point clouds through a decomposition-and-reconstruction strategy. Moreover, it can retrieve semantic information of rooms and doors simultaneously. Another method using 3D primitives can deal with different types of point clouds and retain as much as structural details with respect to protruding structures, complicated ceilings, and fine corners. The experimental results indicate the effectiveness of proposed methods and the robustness against noises and downsampling.
Open Access
A New Cooperative PPP-RTK System with Enhanced Reliability in Challenging Environments
(2023-07) Lyu, Zhitao; Gao, Yang; Wang, Ruisheng; O'Keefe, Kyle; Gao, Yang
Compared to the traditional PPP-RTK methods, cooperative PPP-RTK methods provide expandable service coverage and eliminate the need for a conventional expensive data processing center and the establishment and maintenance of a permanently deployed network of dense GNSS reference stations. However, current cooperative PPP-RTK methods suffer from some major limitations. First, they require a long initialization period before the augmentation service can be made available from the reference stations, which decreases their usability in practical applications. Second, the inter-reference station baseline ambiguity resolution (AR) and regional atmospheric model, as presented in current state-of-art PPP-RTK and network RTK (NRTK) methods, are not utilized to improve the accuracy and service coverage of the network augmentation. Third, the positioning performance of current PPP-RTK methods would be significantly degraded in challenging environments due to multipath effects, non-line-of-sight (NLOS) errors, poor satellite visibility and geometry caused by severe signal blockages. Finally, current position domain or ambiguity domain partial ambiguity resolution (PAR) methods suffer from high false alarm and miss detection, particularly in challenging environments with poor satellite geometry and observations contaminated by NLOS effect, gross errors, biases, and high observation noise. This thesis proposed a new cooperative PPP-RTK positioning system, which offers significant improvements to provide fast-initialization, scalable coverage, and decentralized real-time kinematic precise positioning with enhanced reliability in challenging environments. The system is composed of three major components. The first component is a new cooperative PPP-RTK framework in which a scalable chain of cooperative static or moving reference stations, generates single reference station-derived or reference station network-derived state-space-representation (SSR) corrections for fast ambiguity resolution at surrounding user stations with no need for a conventional expensive data processing center. The second component is a new multi-feature support vector machine (SVM) signal classifier based weight scheme for GNSS measurements to improve the kinematic GNSS positioning accuracy in urban environments. The weight scheme is based on the identification of important features in GNSS data in urban environments and intelligent classification of line-of-sight (LOS) and NLOS signals. The third component is a new PAR method based on machine learning, which employs the combination of two support vector machine (SVM) to effectively identify and exclude bias sources from PAR without relying on satellite geometry. The prototype of the new PPP-RTK system is developed and substantially tested using publically available real-time SSR products from International GNSS Service (IGS) Real-Time Service (RTS).
Open Access
A Study on Efficient Vector Mapping With Vector Tiles Based on Cloud Server Architecture
(2015-12-07) shang, xiaohong; Liang, Steve; Wang, Xin; Wang, Ruisheng; kattan, linda
In web mapping, transmitting large vector data over the Internet has been a challenging issue over the past decade. A method for delivering large vector data in small pieces is known as vector tiling. Generally, studies of traditional vector tile based methods in Web-mapping applications were limited to simple single server-client architecture with GeoJSON encoded vector tiles. However, problems such as limited scalability and inefficient vector tile transmission arose in these studies. To solve these problems, a distributed memory caching implementation has been proposed using cloud architecture. This study also explored the transmission efficiency of three vector tile encoding formats: GeoJSON, TopoJSON, and Google Protocol Buffers. A prototype of the Canada road network vector map was developed. The results of this study show that the proposed solution improves the application performance and is scalable in comparison to naïve architecture.
Open Access
Automated Building Extraction from Remote Sensing Imagery Using Deep Learning
(2022-10-31) Yan, Hailun; Wang, Ruisheng; Bayat, Sayeh; Hay, Geoffrey
Automatically extracting high-quality building-footprint polygons from satellite and aerial images is crucial for supporting various land use and land cover mapping applications. The conventional building polygon extraction process requires hand-crafted features and high human intervention, which is time-consuming and often has limited generalization capability. In recent years, deep learning-based methods have shed light on the problem with higher levels of automation, segmentation quality, and generalization capability. These methods often involve two stages: first, building segmentations are predicted from remote sensing images using deep neural networks; next, the irregular-shaped building segmentations are regularized into straight-edged and right-angle-cornered building polygons using conventional or deep learning-based methods. As a result, the extraction performance is often highly affected by the quality of the segmentation predictions. However, from experiments, the current widely used segmentation DNNs show significant defects in their building segmentation results, especially for the buildings with rotated angles between the building edges and the image edges. Moreover, although DNN-based regularization methods have shown greater generalization potentials at regularizing buildings in various shapes compared to the conventional regularization methods, the qualities of the regularization results are generally dissatisfactory. This thesis proposes an end-to-end deep learning-based building extraction method based on PolygonCNN. The proposed model consists of a segmentation module to predict building segmentations and a regularization module to regularize the building contours traced from the building segmentation results. First, an upgraded Mask R-CNN model, which is integrated with the rotatable bounding box technique, the Swin Transformer backbone network, and the FPN module, is adopted as the segmentation module of the proposed model to segment buildings in vastly different scales and orientations. Moreover, the Feature Pooling module and the BRegNet of the original PolygonCNN are modified to exploit the multi-scale feature maps of the FPN module. As a result, the proposed model can effectively extract high-quality building polygons with various scales and orientations and has shown promising performance compared to several other popular end-to-end deep-learning-based building extraction models. In addition, the thesis provides supplemental architecture choices, which offer flexibility between the quality of the building extraction result and the memory consumption of the model.
Open Access
Automated Recognition of Electrical Substation Components from 3D LiDAR Point Clouds
(2017) Arastounia, Mostafa; Lichti, Derek; Hassan, Quazi; Wang, Ruisheng
This study presents an innovative automated methodology for identification of electrical substations’ key elements from 3D LiDAR point clouds acquired by terrestrial laser scanners. The developed methodology is composed of nine algorithms that identify objects of interest with respect to their physical shape and topological relationships among them. The objects of interest in this contribution are ground, fence, cables, circuit breakers, bushings, bus pipes, insulators, and three types of poles with circular, octagonal, and square cross sectional shape. The developed methodology incorporates a computationally-efficient algorithm for detection of ground within electrical substations; two separate algorithms for identifying well-sampled and poorly-sampled fences; robust algorithms for detecting cables, circuit breakers, and bushings with respect to their unique physical shape and the topological relationships among them; and a novel method for simultaneous identification, modeling, and registration-refinement of poles with circular and regular polygonal cross sectional shapes. The proposed methods in this study work quite robustly despite the challenges introduced by non-uniform point sampling, registration error, occlusion, attached objects, gap, dense configuration of neighboring objects, and outliers. Five datasets with quite different volume and configuration were employed in this work. The first three datasets contain point clouds of two different electrical substations. The fourth and fifth datasets contain point clouds of an urban roadway and a pole-like monument with a regular dodecagonal cross section, respectively. The obtained results indicate that 367 out of 382 objects of interest (96.1%) in the first dataset; 354 out of 382 objects of interest (92.7%) in the second dataset; and 255 out of 264 objects of interest (96.6%) in the third dataset were successfully recognized. At point cloud level, it achieved greater than 99%, 96%, and 97% average recognition precision and accuracy in the first, second, and third dataset, respectively. Furthermore, the poles in the fourth and fifth datasets were successfully identified and the registration-refined version and as-built model of poles in all five datasets were automatically generated. The center and size standard deviation of the constructed models was less than 3 mm and the rotation angle standard deviation was less than 0.3° for all identified poles.
Open Access
Automatic Mapping of Residential Rooftops with High-Resolution Thermal Imagery
(2022-04-25) Ghaffarian, Salar; Hay, Geoffrey J.; Yackel, John; Wang, Ruisheng; Hay, Geoffrey J.;
This study reports on the use of the commercially available ENVI Deep Learning module to (i) automatically extract GIS ready rooftop polygons directly from high-resolution night-time thermal infrared (TIR) airborne imagery and (ii) define the optimal spatial resolution for deep learning rooftop delineation. It also (iii) compares results from multi-spatial resolution models based on a single TIR image vs. a derived three channel image and (iv) introduces two new object-based accuracy assessment methods for comparing the visual fit of the segmented rooftops.
Open Access
Developing the Use of UAV Imagery Systems for Site Specific Weed Management
(2020-09-02) Hassanein, Mohamed; El-Sheimy, Naser; Lari, Zahra; Noureldin, Aboelmagd; Sousa, Mario; Wang, Cheng; Wang, Ruisheng
The use of Unmanned Aerial Vehicle (UAV) imagery systems for Precision Agriculture (PA) applications drew a lot of attention through the last decade. UAV as a platform for an imagery sensor is providing a major advantage as it can provide high spatial resolution images compared to satellite platform. Also, it provides the user with the ability to collect the needed images at any time along with the ability to cover the agriculture fields faster than terrestrial platform. Therefore, these UAV imagery systems are capable to fit the gap between aerial and terrestrial Remote Sensing systems. Weed management is one of the important PA applications that using UAV imagery system for it showed great potentials. The current weed management procedure depends on spraying the whole agriculture field with chemical herbicides to execute any weed plants in the field. Although such procedure seems to be effective, it has huge effect on the surrounding environment due to the excessive use of the chemical, especially that weed plants don’t cover the whole field. Usually weed plants spread through only few spots of the field. Therefore, different efforts were introduced to develop weed detection techniques using UAV imagery systems. Though the different advantages of UAV imagery systems, such systems didn’t draw the users interest due to many limitations such as the cost of these systems. The primary objective of the research work is to develop the use of UAV imagery systems for PA with focus on weed management through tackling the different limitations of using UAV imagery systems for weed management. Therefore, different methodologies are introduced for vegetation segmentation, crop row detection, and weed detection. These methodologies are able to enhance the use of low-cost UAV imagery systems through targeting two main goals. First, the use of RGB imagery sensors. Second, collect the imagery data from high altitudes.
Open Access
External and Internal Corrosion and Its Control of Natural Gas Pipelines
(2019-12) Qian, Shan; Cheng, Y. Frank; Li, Leping; Wang, Ruisheng; Oguocha, Ikechukwuka; Wong, Ron
Natural gas pipelines suffer from both external and internal corrosion during their service life, which may result in dramatic consequences. In this research, external corrosion of X52 pipeline steel under direct current (DC) interference was investigated in a simulated soil solution. Corrosion acceleration by DC was quantitatively determined as a function of DC current density. DC was found to shift the cathodic protection (CP) potential to positive and negative directions in the anodic and cathodic zones, respectively, on the pipelines, resulting in either corrosion enhancement or hydrogen evolution at the zones. The effect of DC on properties and performance of fusion bonded epoxy (FBE) coating applied on pipelines was studied. The presence of DC interference facilitates water permeation into the coating due to the altered molecular structure and decreased the coating resistance for corrosion protection. Furthermore, corrosion of X52 pipeline steel under dynamic DC interference was investigated. Dynamic DC further accelerates the steel corrosion compared to static DC at specific DC current densities. It is believed that the alternating current (AC) component included in the pulse DC contributes to the corrosion reaction. With increase in the DC pulse frequency, corrosion rate of the steel decreases. The wave form of the dynamic DC does not obviously affect the steel corrosion. Internal corrosion of X52 pipeline steel was investigated in CO2-containing thin layers of solution, simulating the actual corrosive environment generated in the interior of natural gas pipelines. A mechanistic model was developed to explain the internal corrosion of wet gas pipelines. With the decrease of the solution layer thickness, the corrosion rate of the steel reduces. An elevated temperature accelerates the corrosion reaction kinetics, and generates a compact and homogeneous FeCO3 film at the same time. The presence of acetic acid increases the steel corrosion, while the methanol reduces corrosion rate of the steel. For external corrosion control, a micro/nanostructured ZnO-alkylamine composite coating was developed by electrodeposition and anodization to possess multiple functions. The optimal coating film is superhydrophobic, with the water contact angle up to 158o. The coating possesses a good corrosion resistance and excellent self-cleaning performance and a strong anti-adhesion to pseudomonas aeruginosa bacteria. For internal corrosion control, the inhibition performance of imidazoline (IM) and sodium dodecylbenzenesulphonate (SDBS) inhibitors and their synergism on corrosion of X52 steel in CO2-saturated chloride solutions was investigated. The synergistic effect of the two inhibitors enhances the corrosion inhibition performance, compared to the inhibitors acting independently. The adsorption of both inhibitors on the steel is chemisorption, following the Temkin adsorption isotherm.
Open Access
Fully Convolutional Networks for Semantic Segmentation of Very High Resolution Remotely Sensed Images
(2018-04-24) Sun, Weiwei; Wang, Ruisheng; Wang, Xin; Lichti, Derek D.
The semantic segmentation of very high resolution (VHR) remotely sensed images is to assign a categorical label for each pixel, which is an important but unsolved problem in remote sensing. In recent years, fully convolutional networks (FCN) have become the state-of-the-art framework for the semantic segmentation in computer vision. Thus, this work aims to improve the semantic segmentation of VHR images by utilizing FCN. Firstly, we propose a promising framework which achieves the top result (90.6%) on the ISPRS Vaihingen benchmark. In the framework, the proposed FCN-based network obtains a competitive result (90.1%). In addition, we develop the DSM backend to enhance the result of FCN by incorporating complementary information from color images and digital surface model (DSM). Secondly, we propose the recurrent FCN for modeling the continuous context inherent in VHR images. Experimental results demonstrate that the recurrent FCN significantly boosts the performance of FCN by incorporating the local contextual information from patches and the global contextual information between patches.
Open Access
Geometric Primitives in MLS Point Clouds Processing
(2020-04-14) Xia, Shaobo; Wang, Ruisheng; Lichti, Derek D.; Shahbazi, Mozhdeh M.; Gao, Yang; Kang, Zhizhong
Mobile Light Detection and Ranging (LiDAR), as an active remote sensing system, has become an accessible street-level mapping technology in the last decade due to its ability to collect accurate and dense 3D point clouds efficiently. Although tremendous effort has been made to LiDAR data processing, there still exist many problems in everyday tasks ( e.g., segmentation and detection). In this thesis, the LiDAR data processing is re-visited from a geometric-primitive perspective, with the hope that existing problems can be partly solved or even well addressed by tapping the potential of geometric primitives. A survey on geometric primitive extraction, regularization and their applications is presented for the first time. In this review, geometric primitives that consist of a group of discrete points are categorized into two classes: shape primitives (e.g., planes) and structure primitives (e.g., edges). The rest of this thesis focuses on geometric primitives in mobile LiDAR data processing. A fast 3D edge extraction method which consists of finding and linking edge candidates is proposed and tested in large-scale scenes. Given extracted edge clusters, a new facade separation method for mobile LiDAR point clouds is developed, based on which connected facades are separated into facade instances for the first time. To explore the potential of plane primitives in mobile LiDAR data processing, a novel instance-level building detection method based on plane primitives extracted from original point clouds is proposed. After that, a new point cloud segmentation algorithm that succeeds in separating buildings and vegetations is presented. The main contribution lies in using plane priors to improve segmentation accuracy. For line primitives, a new extraction method is presented in this thesis, which can extract multiple primitives simultaneously from projected point clouds. Based on extracted line segments, a graph-based method is presented to construct 2D building footprints. Last but not least, this thesis also introduces the energy-based ``hypothesis and selection" (HS) framework to object detection and segmentation in LiDAR point clouds for the first time. The adapted frameworks are proved to be flexible and effective according to extensive experiments in different applications.
Open Access
Geometry and radiometry based enhancement for 3D reconstruction from imagery
(2020-05-01) Mohammed, Hani Mahmoud Mohammed; El-Sheimy, Naser; Noureldin, Aboelmagd M.; Wang, Ruisheng
3D reconstruction from imagery is making significant contributions in a wide array of applications, including but not limited to heritage documentation, 3D modelling of the human body, and digital elevation model (DEM) generation. The importance of imagery-based 3D reconstruction is a result of being a low-cost alternative to active sensors, such as laser scanners. However, it is unfortunate that the point cloud generated from active sensors most likely will be more accurate than a point cloud generated from stereo imagery or structure from motion (SFM). Besides the scale ambiguity, which can be resolved using the associated navigation sensors like the inertial measurement unit (IMU) and the GPS receiver, there are other issues that limit the quality of imagery-based 3D reconstruction. Reconstruction from imagery involves several stages; and in most of these stages, the outcome is based on estimation procedures, which means that the whole process is more probabilistic than deterministic. Furthermore, noise and radiometric errors in the input images can lead to errors in feature detection and matching and can dramatically affect the quality of the orientation solution. The effects of noise and radiometric errors are not limited to the deterioration of the orientation parameters, however, and also can have a significant impact on dense matching, the stage at which the point cloud is generated. In this dissertation, an enhancement workflow for imagery-based 3D reconstruction is proposed. The proposed workflow consists of three approaches, each of which tackles one common problem in the process of 3D reconstruction. The first approach aims at enhancing the feature detection and matching in image pairs in which both the geometric and radiometric information of the image pairs are fused to find an accurate and robust set of matches. In the proposed approach, a small subset of matches is used to estimate an initial solution for the homography and fundamental matrix. These two entities constrain the point correspondence to a specific area. Then, a radiometric correlation is applied to find the desired set of matches in a recursive way. In the second approach, the noise in disparity maps is removed while image segmentation is being performed at the same time. The main target of the proposed approach is to segment both the disparity map and the original images based on both the geometric and radiometric constraints. First, the disparity map is roughly segmented using its grey-level histogram and basic region growing/thresholding algorithms. Then, a homography relation between the image pairs, based on the segmented disparity map, is constructed. This allows labelling each connected region in the image pair as a segment. The segmentation process is further enhanced using colour edge detectors and spatial and frequency domain filters. The third approach enhances the dense matching process using the segmented disparity map. The segmented disparity map obtained from the second approach is considered to be almost noise-free as a result of applying the homography-based segmentation and that all invalid disparity points have been detected and rejected. Therefore, the proposed approach aims to replace those points with more accurate points based on the local homography and the colour constraints imposed on the image pair. This approach is considered a post-processing approach, and it is therefore assumed that the disparity map is already known. The outcome of the proposed approaches is a segmented, more accurate, and well-defined 3D point cloud with less noise compared to the current 3D reconstruction approaches.
Open Access
Improving the Accuracy of GNSS Receivers in Urban Canyons using an Upward-Facing Camera
(2018-07-03) Gakne, Paul Verlaine; O'Keefe, Kyle P. G.; Gao, Yang; Wang, Ruisheng; Fapojuwo, Abraham Olatunji; Ruotsalainen, Laura
Global Navigation Satellite Systems are widely used as localization systems for various applications in indoor and outdoor environments. Autonomous vehicles for example rely on navigation sensors such as GNSS receivers, INS, odometers, LiDAR, radar, etc. However, none of these sensors alone is able to provide satisfactory position solutions in terms of accuracy, availability and reliability all the time and in all environments. This thesis presents a new tightly coupling method fusing the egomotion of a land vehicle estimated from a sky-pointing camera with GNSS signals and a digital map for navigation purposes in harsh urban canyon environments. The advantages of this configuration are three-fold: firstly, for the GNSS signals, the upward-facing camera will be used to classify the acquired images into sky and non-sky (known as segmentation). A satellite falling into the non-sky areas (e.g., buildings) will be rejected and not considered for the final position solution computation. Secondly, the narrow field of view sky-pointing camera is helpful for urban area egomotion estimation in the sense that it does not see most of the moving objects (e.g., cars) and thus is able to estimate the egomotion with fewer outliers than is typical with a forward-facing camera. Thirdly, the skyline can be extracted and serves as a finger print of the vehicle location in the city. This information can then be correlated with a 3D city model to obtain the vehicle location. In order to obtain an accurate solution from the proposed method, a few intermediate steps had to be taken into account. An improved image segmentation algorithm is presented. The output of this algorithm served for the skyline positioning and the camera-based multipath mitigation. Also, an accurate visual odometry was implemented. Moreover, the monocular-based visual odometry is able to determine the vehicle translation accurately but up to a scale only. An integrated system that tackles the scale factor issue is designed. From five datasets evaluated in this research, the proposed method has shown to be robust and provide more accurate position, velocity and attitude solution at least 83% of the time than the GNSS-only and loosely coupled GNSS/vision solutions.
Open Access
A Model-based, Optimal Design System for Terrestrial Laser Scanning Networks in Complex Sites
(2019-08-29) Jia, Fengman; Lichti, Derek D.; O'Keefe, Kyle P. G.; Wang, Ruisheng; Shahbazi, Mozhdeh M.; Lindenbergh, Roderik C.
With the rapid increase of terrestrial laser scanner (TLS) applications, especially for the high-accuracy modelling of large-volume, complex objects, a design system is required to provide the optimal solutions for both scanner and target placement, so that the project requirements in terms of coverage, precision, economy and reliability can be met. In this thesis, a model-based, optimal design system for terrestrial laser scanning networks in complex sites is proposed. First, a hierarchical TLS viewpoint planning strategy driven by an improved optimization method is developed to solve the optimal scanner placement problem. The main contribution of the proposed method is to improve the efficiency in design without jeopardizing the optimality of the solution, compared with the traditional method with the extensive search strategy. In addition, the target placement for registration, which draws limited attention in the existing research, is determined by optimizing the target arrangement criterion, and the number of target locations is minimized by accepting the close to optimal target arrangement. Finally, the quality of the design, including the sensitivity of the object coverage to viewpoint placement and the precision of the point cloud are provided. The proposed methods were verified by the relatively small network first and then applied on two building complexes located on the University of Calgary campus. The design for scanner placement was compared with the “brute force” strategy in terms of the optimality of the solutions and runtime. The results showed that the proposed strategy provided scanning networks with a compatible quality but a significantly improved efficiency in design. The number of target locations necessary for registration from the proposed system was surprisingly small, considering the volume and complexity of the networks. Through the quality assessments, the sensitivity of the object coverage to the scanner placement indicated where users might need to consider viewpoint densification, and the point cloud precision indicated if the network design could meet the project requirements.
Open Access
Modeling the Loading and Fate of Estrogen
(2015-12-04) Fleming, Michael; Achari, Gopal; Hassan, Quazi; He, Jennifer; Wang, Ruisheng
Endocrine disrupting compounds may produce infertility, nervous system disorders, and improper functioning of the immune system in humans and wildlife. Estrogens are classified as the most potent and common endocrine disrupting compounds, and the major point source for estrogen is municipal wastewater. Monitoring of estrogen is challenging, expensive, and intermittent; and therefore, the focus of this work is modeling estrone, 17β-estradiol, and 17α-ethynylestradiol concentrations from wastewater treatment plants in Calgary and Edmonton, Alberta, and Brandon, Manitoba. Demographic groups, excretion rates, population estimates, average daily flows, calculated estrogen transformation, calibration, calculated influent-to-effluent reduction percentages, and a treatment unit removal matrix are used to determine loading estimations of estrogen. The results demonstrate reasonable accuracy against previous measurements, and findings are consistent with concentrations reported in the literature. Upon further calibration with additional local data, the model may be used as a risk assessment analysis tool for these contaminants of concern.
Open Access
Multi-Sensor Map Matching Techniques for Autonomous Land Vehicle Navigation
(2016) Balazadegan Sarvrood, Yashar; Gao, Yang; Wang, Jinling; El-Sheimy, Naser; Wang, Ruisheng; Helaoui, Mohamed
This thesis proposes a method for tight integration of digital map and multi sensors including Dead Reckoning (DR) system and Precise Point Positioning (PPP). First, the digital map is tightly coupled with the DR system, including stereo Visual Odometer, Light Detection And Ranging (LiDAR) Odometer and reduced Inertial Measurement Unit (IMU), including two horizontal accelerometers and one vertical gyro. The algorithm starts with stereo Visual Odometry to estimate six Degree of Freedom (DoF) ego motion including rotation and translation parameters to register the point clouds from previous epoch to the current epoch. Afterwards, a Generalized Iterative Closest Point (GICP) algorithm is used to refine the stereo Visual Odometry motion estimation. Then, an Extended Kalman Filter (EKF) is used to integrate the forward velocity and azimuth obtained by Visual-LiDAR Odometer and reduced IMU outputs to provide the final navigation solution. This integrated navigation solution is the input to the fuzzy logic based Map Matching (MM) algorithm, which takes the imprecise and noisy inputs and gives the crisp outputs. The fuzzy logic MM goal is to identify the correct road link, and to determine the vehicle location on the selected road link. The proposed fuzzy logic MM consists of two distinct steps: 1) The Initial Map matching Process (IMP) and 2) The Subsequent Map matching Process (SMP). The proposed map matching algorithm improves integrated multi sensors (stereo Visual-LiDAR and reduced IMU) position accuracy by constraining the vehicle location on the road. The map matching provides close-loop controls for the Dead Reckoning (DR) drift errors by feeding back the map matched position and road link azimuth to the reduced IMU mechanization. This research proposes a new software system for tight integration of kinematic PPP and digital map as well. The PPP provides the navigation solution for MM and MM finds the correct road link and improves PPP performance by providing the map matched position and link azimuth as feedbacks to the Kalman Filter (KF) of PPP. In this research two datasets were used. 1) The datasets from KITTI (Karlsruhe Institute of Technology and Toyota technological Institute) to tightly couple digital map and integrated stereo Visual-LiDAR and reduced IMU, 2) The datasets collected by Positioning and Mobile Information System (PMIS) Group at University of Calgary to tightly couple digital map and integrated stereo Visual Odometry (VO) and reduced IMU and to tightly couple kinematic PPP and digital map. The results show that Visual Odometry (VO)-LiDAR is more accurate than Wheel Odometer, because it provides azimuth aiding to vertical gyro, resulting in a more reliable and accurate system. A low-cost system is developed by using two cameras plus reduced IMU. The cost of such a system will be reduced than using full tactical MEMS (Micro-Electro-Mechanical Sensor) based IMUs, because two cameras are cheaper than full tactical MEMS based IMUs. The results indicate that integrated stereo Visual-LiDAR Odometry and reduced IMU can achieve accuracy at the level of the state of the art. Moreover, tight integration of digital map and integrated stereo Visual-LiDAR Odometry and reduced IMU can achieve considerably better accuracy than existing methods. Moreover, tight integration of digital map/DR gives considerably higher correct link identification rate and lower Root Mean Square Error (RMSE) than a loose integration of digital map/DR. In addition, tight integration of digital map and kinematic PPP outperforms stand-alone PPP and reduces the horizontal RMSE and the convergence time of the float ambiguities.
Open Access
Multiuser Usability of Collaborative Virtual Environments
(2017) Erfanian, Aida; Hu, Yaoping; Far, Behrouz H.; Yanushkevich, Svetlana; Wang, Ruisheng; Latoschik, Marc E.
Collaborative virtual environments (VEs) require suitable interaction models for resolving conflicts and promoting multiuser usability. An interaction model is a key component of a collaborative VE. Traditional models such as the first-come-first-serve (FCFS) model have a problem of disregarding the vital socio-human need of equality in interaction (i.e., EII). This problem may impair the suitability of a model. Other components of a collaborative VE, including interaction devices and communication cues, may also affect the suitability of a model. Common cues are verbal and vibrotactile cues. Traditional usability studies on collaborative VEs suffer from several shortcomings. First, a set of multiuser usability metrics are not defined to consider socio-human needs and cover all possible usability factors presented by recent international standards. Secondly, suitable models to address these needs have not been sufficiently investigated. Finally, there have been a lack of studies that investigate the role of devices and cues on the suitability of models. To address these shortcomings, this thesis proposes a framework of multiuser usability for assessing collaborative VEs. The proposed framework consolidates socio-human needs and standard factors of usability. Moreover, a dynamic priority (DP) model that considers the vital need of EII is proposed to address the shortcomings of traditional models. The proposed DP model grants interaction opportunities to users based on the recency of their gained accesses. Investigations under the proposed framework indicated that compared to the FCFS model, the DP model yields perceived EII independent of devices and significantly improves the multiuser usability. The DP model also yields perceived EII regardless of cues. However, a combination of verbal and vibrotactile cues significantly promotes the multiuser usability of a VE governed by the DP model. These results imply the suitability of the DP model as well as combined verbal and vibrotactile cues to promote the multiuser usability within VEs.
Open Access
Neural Representation for 3D Building Reconstruction from Point Clouds
(2024-09-19) Akwensi, Perpetual Hope; Wang, Ruisheng; Hassan, Quazi Khalid; Ioannou, Yani Andrew; Yang, Hongzhou
The continuous rise in urban growth has underscored the importance of airborne LiDAR point clouds (APCs) for efficient/cost effective urban planning, management, and development. However, the delineation and modeling of 3D objects -- specifically buildings -- from APCs pose significant challenges due to issues such as façade/roof occlusions, point density variations, sparsity, and noise. This thesis aims to address these challenges and provide neurally-driven solutions for 3D digital twinning of buildings from APCs. To delineate an urban scene into object categories, PReFormer, a memory-efficient point transformer capable of achieving competitive segmentation results with fewer model parameters and less memory is proposed. The PReFormer comprises of an optimized point embedding module, linearized multi-head self-attention layers, and reversible functions, all designed to reduce computation time and space complexities. Additionally, the architectural design of the PReFormer follows a ∇-shape, which improves (object) size-invariant feature extraction and segmentation accuracy. To generate high fidelity 3D building models from delineated APC building instances, APC2Mesh, a framework which integrates 3D building point completion and reconstruction processes, is proposed. The developed point completion network uses dynamic edge convolution and self-attention mechanism operations to extracts both local and global building shape information for complete building point set (BPS) reconstruction. This completion process mitigates the sparsity, occlusion, and point density variability issues usually associated airborne LiDAR BPSs. Leveraging the completed BPSs, a linearized skip-attention-based deformation network capable of handling several building styles and/complexities is presented to generate high fidelity 3D building mesh models. An observation of the mesh models from APC2Mesh shows that mesh models have relatively high disk storage, and are difficult to manipulate given their numerous triangular mesh faces. Thus, Points2Model, a neural-guided method that reconstructs building wireframes from APCs is further proposed. It uses neural implicit learning to up-sample completed BPSs, and a simple yet robust corner-focused hypothesis and selection strategy to detect building corners and their corresponding edge connectivity. Overall, this thesis presents innovative solutions for overcoming the inherent challenges of APCs in 3D building reconstruction, thus contributing significantly to the field of digital twinning of urban buildings.
Open Access
Personalized Travel Route Recommendation Based on GPS Trajectories
(2018-06-28) Cui, Ge; Wang, Xin; Zhong, Ming; Wang, Ruisheng; Chen, Zhangxing; Liang, Steve H. L.; Rangelova, Elena V.
Travelling is a critical component of daily life. With new technology, personalized travel route recommendations are possible and have become a new research area. A personalized travel route recommendation refers to plan an optimal travel route between two geographical locations, based on the road networks and users’ travel preferences. In this thesis, it first proposes a segment-based map matching method to locate GPS trajectories onto the road network, and then extract users’ travel behaviours from their historical routes. Next, users’ travel behaviour frequencies are estimated by using collaborative filtering technique. This thesis defines two types of travel behaviours, appearance behaviour and transition behaviour, from users’ historical GPS trajectories and propose three personalized travel route recommendation methods¬, including CTRR, CTRR + and Map2R, to consider users’ personal travel preferences based on their historical GPS trajectories. A route with the maximum probability of a user’s travel behaviour is then generated. CTRR only considers user’s appearance behaviour and calculates the maximum probability route based on naïve Bayes model. Besides, CTRR is extended to CTRR+ by integrating distance with the user appearance behaviour probability. In MaP2R, it considers both appearance behaviour and transition behaviour, and calculate the maximum probability route based on Markov model. This thesis also conducts some case studies based on a real GPS trajectory dataset from Beijng, China. The experimental results show that the proposed CTRR methods achieve better results for travel route recommendations compared with the shortest distance path method, and both CTRR+ and MaP2R can enhance the performance of CTRR, respectively.
Open Access
Photogrammetric Modelling for 3D Reconstruction from a Dual Fluoroscopic Imaging System
(2019-01-03) Al Durgham, Kaleel Mansour; Lichti, Derek D.; Kuntze, Gregor; Wang, Ruisheng; Shortis, Mark R.; Boyd, Steven Kyle; Ronsky, Janet L.
Biplanar videoradiography (BPVR), or clinically referred to as dual fluoroscopy (DF), imaging systems are increasingly being used to study the in-vivo skeletal biomechanics of human and animal locomotion. DF imaging provides a novel solution to quantify the six-degree-of-freedom (6DOF) skeletal kinematics of humans and animals with high accuracy and temporal resolution. Using low-dose X-ray radiation, DF systems provide accurate bone rotation and translation measurements. In this research domain, a DF system comprises two X-ray sources, two image intensifiers and two high-speed video cameras. The combination of these elements allows for the stereoscopic imaging of the bones of a joint at high temporal resolution (e.g., 120-250 Hz), from which bone kinematics can be estimated. The utilization of X-ray-based imaging results in challenges that are uncommon in optical photogrammetry. Unlike optical images, the inherent lack of colour information in DF images complicates fundamental tasks such as the derivation of image observations for the system calibration. Furthermore, the incorporation of an image-intensifier to produce DF images results in high distortion artifacts that are uncommon in optical photogrammetry. The use of image intensifiers also results in non-uniform intensity response in the DF images. Unlike optical images with well-established camera models, the systematic distortion behaviour in DF images is empirically modelled. The novelty in this research work is in providing a complete, scientific, straightforward and accurate photogrammetric framework for deriving 3D measurements from a DF imaging system. This research work provides means for automating the DF calibration procedure and introduces solutions for improving the methodology of 3D reconstruction from DF imaging. A thorough photogrammetric analysis over the system aspects points out the weaknesses in the iii traditional 3D reconstruction procedures and suggests accurate alternatives. The dissertation presents five scientific contributions: (1) a semi-automated methodology to derive the image observations from time series DF calibration images, (2) validation of an empirical DF sensor model (bundle adjustment-based) for the calibration of the DF system and introducing it as a superior replacement for the traditional direct linear transformation-based (DLT) calibration approaches, (3) a rigorous accuracy assessment methodology for the evaluation of the DF system reconstruction capabilities, (4) a novel methodology for the temporal stability analysis of an imaging system calibration parameters, and (5) a virtual-3D-model means to facilitate establishing the alignment between stereoscopic DF image pair and an MRI/CT model (2D-to3D registration).