About GEG · GEG Projects · GEG Publications & Presentations
Good Governance of Canada's Oceans
Current and Recently Completed GEG Projects
Accuracy Assessment of Orthoimages Derived from High Resolution Satellite Imagery
H. Afify and Y. Zhang
The objective of this study was to evaluate the geometric accuracy of orthoimages derived from high-resolution imagery. The Rational Function Model (RFM) with different orders was applied to the autorectification. Ikonos and QuickBird panchromatic images covering Fredericton, New Brunswick, Canada were tested. The terrain-dependent scenario was considered, for which the Rational Function Coefficients (RFCs) were derived using different numbers of Ground Control Points (GCPs). Quantitative evaluations of the geometric accuracy of the orthoimages were performed using independent sets of checkpoints. The results show the potential use of different RFM orders with user-derived coefficients and different numbers of GCPs for the improvement of geometric accuracy of orthorectification of Ikonos and QuickBird images.
Detection of the Land-Cover Changes in Fredericton Using Landsat TM and Ikonos Imagery
S. Burbridge and Y. Zhang
The purposes of this research project were (1) to examine the extent to which high-resolution change detection can be performed using a combination of high and medium-resolution satellite imagery, and (2) to provide some relevant land cover change information to the City of Fredericton. Different change detection techniques were evaluated. The post-classification change detection algorithm combined with neural network classification approach was selected for the research. A Landsat 5 TM image set, 1994, and an Ikonos MS image set, 2000, of Fredericton were employed in the change detection. Successful urban land cover changes were monitored to an acceptable degree of certainty. However, some problems in terms of detailed detection, such as separation of individual family houses from small streets, still remain. Further research on these problems is ongoing.
[DND]
Virtual Tunnel: A Spatial Information System for Tunnel-Type Network by Integrating GIS and Image-Based Rendering (IBR) Techniques
Richard Chan, M.Sc.E. Candidate; David Coleman
Panoramic images are used widely to simulate a 'be-there' experience by allowing user to select one's viewpoint at a fixed location. Recently, more and more research has been carried out on using panoramic images to create photo-realistic virtual reality that allows the user to navigate along a pre-defined path or within a certain space limit. These images are usually created by capturing a number of images that cover the 360° horizontal view around a vertical axis. They are then re-sampled to a cylindrical projection and stitched to form a large "horizontal panoramic image".
Similarly, "vertical panoramic images" can also be created to form a 'view tunnel' that capture the 360° vertical view along the path inside a tunnel. By overlapping the vertical panoramas, stereo pair of any part of the internal surface of a tunnel can be created for photogrammetric operations. They can also be used as a backdrop for spatial data visualization, especially for 3-dimensional data visualization.
This research is intended to demonstrate a new way in creating a photo-realistic virtual environment by using a series of overlapping vertical panoramas. A prototype is to be developed to: (1) provide a data model to organize spatial data of a tunnel-type network system, including profiles and image data; (2) provide a visualization system to allow user to navigate along the tunnel to inspect the spatial data in a photo-realistic virtual reality environment; and (3) provide stereo view of the internal surface of the tunnel for various data processing activities such as measuring, labeling, drawing or photogrammetric operations.
Evaluation of State-of-the-Art Polarization Technologies for Accuracy Improvements of SAR Image Classification
C. Deschênes and Y. Zhang
Open source and commercially available software packages for SAR polarization processing were tested and evaluated using simulated Radarsat II quadrature polarization data. A variety of polarization results with specific characteristics were obtained and then incorporated into the classification for different land covers. The evaluation results demonstrate that the polarization information in a specific direction is beneficial for the classification of specific land cover classes.
[DND, PCI Geomatics, Canadian Space Agency]
Integration of Sea Ice Information Into ECDIS
George Dias, M.Sc.E. Candidate; David Coleman; Ahmed El-Rabbany (Ryerson/Crestech)
Safe and efficient marine navigation in ice-infested waters require comprehensive and timely information on the sea ice conditions. The ice charts produced daily by the Canadian Ice Service enable mariners to make critical decisions regarding the selection of the best possible navigation routes.
While highly useful in providing the mariners with comprehensive ice information, these ice charts may not fulfill the requirements for safe and efficient marine navigation, even when they are used side-by-side with Electronic Chart Display and Information Systems (ECDIS). Funded by CresTech and completed in collaboration with Ryerson University this research proposes that, in ice-infested waters, an integrated navigation chart system may be developed which integrates vital ice information into an ECDIS in a formal manner and in accordance with International Hydrographic Organization standards. In addition, a modified production workflow is being developed and tested that would optimize production from initial satellite image acquisition through creation of digital, ECDIS-ready ice information.
Calibration of LIDAR Sensor
S. Ghayourmanesh and Y. Zhang
We joined an industry research project to calibrate a commercially available LIDAR sensor system for precise elevation measurements. The existing algorithms and software for distance calculation were improved based on the calibration parameters, resulted in significant improvement of elevation measurements.
[Terra Remote, Vancouver, BC]
DEM Generation Using a Single Set of SAR Polarization Data
S. Ghayourmanesh and Y. Zhang
In this project, Shape from Shading (SFS) techniques have been significantly improved to extract elevation information from a single set of SAR data. New techniques involving full polarization information of potential Radarsat II data will be developed to further improve the accuracy of elevation information extraction. If successful, this will be a breakthrough technology to extract elevation information from a single set of SAR polarization data.
[Canadian Space Agency, NBIF]
Financing Models for Spatial Data Infrastructure and Their Applicability to Emerging Nations
Garfield Giff, Ph.D. Candidate; David Coleman
National spatial data infrastructure (SDI) initiatives now in place are all aimed at improving the quality and availability of spatial information throughout their respective constituencies. The majority of this first generation of SDIs evolved from National Mapping Agencies and thus, a significant proportion of their funding was derived from the budgets of these organisations. Support for their implementation was also derived from funds set-aside by government and international funding agencies for special projects.
Current SDIs are now maturing and proving to be an integral part of the information society supplying spatial information of potential significant economic value [Andari Consultants, 2001]. However, the structured financing mechanism proposed may not be suitable for ultimate restructuring, implementation and maintenance of future SDIs. Moreover, these models were designed and developed for the implementation environment of developed nations and may not be applicable in emerging nations.
This research compares the financing of spatial data infrastructure with earlier types of infrastructure, and then examines alternative models for financing and implementation. After analyzing the general implementation environments of emerging nations and determining their effect on the proposed models, the research will then customize these models for application in the relatively different economic and political environments of emerging nations.
Spatial Contiguity-Constrained Clustering by Expectation Maximization
Aimin Gong, Ph.D. Candidate; Francois Anton; David Coleman
In many GIS applications, it is often necessary to partition a set of N multivariate observation vectors (called attributes) located at neighboring geographic areas, into K groups (called regions) so that the areas within a region are not only similar to each other in term of the attributes but also geographically connected. But each existing method for contiguity-constrained clustering has its weakness. Further, while heuristics and criteria like minimum description lengths (MDL) and Bayesian Information Criterion (BIC) have been proposed for selecting the number of clusters in a dataset, previous research efforts have rarely discussed this problem in the case of clustering of spatial data.
The objective of this research is to find more efficient and effective methods for contiguity-constrained clustering that overcome some of the disadvantages of existing methods such as the chaining effects of single-link clustering methods. Methods will be based on mixture model and the Expectation Maximization (EM) algorithm, combining ideas from traditional hierarchical clustering methods. After tentatively determining and testing the number of components in clustering of spatial data under contiguity-constraints, the proposed clustering method will be implemented and apply to Statistics Canada census datasets in order to build specific "communities" which best satisfy certain demographic criteria.
Colourization of SAR Images Using Low Resolution Multispectral Images
G. Hong and Y. Zhang
We developed a software tool for colourization of SAR images using low resolution multispectral images, which has been licensed to Intermap Technologies for value added production. The software tool was developed based on the wavelet integrated image fusion algorithms developed in our research project funded by GEOIDE.
[Intermap Technologies]
Algorithm and Software Development for Fast Detection of Environmental Changes Using Ikonos and QuickBird Images
G. Hong and Y. Zhang
This research is exploring a fast and cost effective way to detect environmental changes using original and pan-sharpened high-resolution multispectral satellite images, such as Ikonos and QuickBird. The research is focused on the following four aspects: (1) radiometric correction, (2) automatic image registration, (3) change detection, and (4) accuracy assessment. Some initial promising results, such as wavelet and IHS integrated fusion, have been achieved. Further research is ongoing.
[GEOIDE, NSERC]
Performance Evaluation of Change Detection Algorithms and Software Packages
G. Hong and Y. Zhang
The performances of latest algorithms and software packages for change detection were evaluated in this project. Change detection algorithms published in literatures were reviewed. Latest commercially available software packages were evaluated. Different high resolution airborne and satellite-borne optical images were used in the evaluation to detect changes of different man-made objects. The most effective change detection algorithms and software packages were recommended.
[DND]
Evaluation of Pan-Sharpening Techniques for the Improvement of Classification Accuracy
T. Maxwell, K. Monroe and Y. Zhang
A variety of commercially available image fusion software tools were employed to fuse panchromatic and multispectral images from the latest satellites such as Ikonos, QuickBird and Landsat 7. Different classification methods, including per-pixel based classifiers and object-oriented classification tools, were applied to the classification of original multispectral images and pan-sharpened multispectral images. Comparison results proved that the PCI Pansharp (an automatic fusion technique developed by Y. Zhang) can improve the classification accuracy, especially when object-oriented classification is employed; while other fusion techniques resulted in inconsistent classification results.
[NSERC, DND]
Supervised Segmentation for Object-Oriented Classification
T. Maxwell and Y. Zhang
Object-oriented classification has proved to be a promising direction for classifying high resolution multispectral images. Image segmentation is a crucial component of object-oriented classification. However, state-of-the-art technologies rely on trail and error methods to obtain reasonable segmentation results, which is a tedious and operator dependent process. In our research, a supervised segmentation technique was developed to obtain optimum segmentation results by training the system using a few selected objects. Significant improvements in process efficiency and result accuracy were achieved and the results are operator independent.
[DND, PCI Geomatics]
Semi-Automatic System for Building Extraction from QuickBird Images
D. Mayunga, D. Coleman, and Y. Zhang
A semi-automatic system has been developed to extract buildings or houses in informal settlements from QuickBird panchromatic images. Algorithms and software based on Snakes techniques have been developed to extract individual house outlines semi-automatically, starting from a seed point inside of the roof of a house. Significant improvements have been achieved compared to existing Snakes based techniques.
[Commonwealth Scholarship]
Data Structures and Algorithms for the Integration of Raster and Vector GIS
D. Mioc, O. Sharma and G. Dharmaraj
Data capture and conversion are two of the most costly operations of any GIS, in terms of computer time and work needed for spatial data acquisition. They can represent up to 80% of the total implementation costs. Fast data acquisition from raster data sources like scanned maps and satellite imagery could significantly improve the operation of GIS. Currently, commercial GIS do not offer completely automatic raster/vector conversion even for simple scanned black and white maps.
The main objective of this research is to use the topological approach to develop new algorithms and data structures for integrated raster/vector models leading to the improvement of data capture and conversion in GIS.
The methodology is based on image processing techniques to extract the basic spatial features from raster data. These spatial features can be used for the reconstruction of the image within the topological data structure - the Voronoi diagram. Voronoi diagrams are irregular tessellations of space, that adapt to spatial objects and therefore they are a synthesis of raster and vector data models. The novel part of this research is the definition of deterministic topological rules and algorithms for extracting the spatial features from the Voronoi data structure.
[NSERC]
Gender Issues and Best Practices in Land Administration
Sue Nichols, Renee Gioverelli, Susana Lastarria-Cornhiel, and Beth Katz
A comprehensive multi-disciplinary review of four World Bank land administration projects in Ghana, Azerbyjan, Bolivia, and Laos was undertaken to determine whether the projects had impacts on women's access to land and what lessons could be learned from each project.
[World Bank]
Women's Access to Land
Sue Nichols, Katie Komjathy, and Hazel Onsrud
This research is on-going and has included:
- development and publication of gender guidelines for land administrators
- an analysis of the impact of land reform in Kyrgystan on women and the poor
- seminars and research on gender issues in China, Vietnam, and Russia
- development of guidelines for NGOs
- demonstration projects and seminars in Brazil
Design of a Marine Cadastre for Musquash Marine Protected Area
Sue Nichols, Sam Ng'ang'a, John Hughes Clarke, Dave Monahan, and Ted Byrne
Based on field data collected during the ocean governance project, the research explored the concept of a marine cadastre to support development marine protected areas.
[GEOIDE, DFO, SNB, NSERC]
Land Policy and Administration
Sue Nichols, Boipuso Nkwae, Mele Rakai, Erik Holmlund, Hazel Onsrud, and Meredith Hutchison
This research involved a number of projects including:
- improving peri-urban land tenure and administration in southern Africa
- capacity building in land administration in Cambodia, southern Africa, and Eastern Europe
- appropriate land tenure and administration systems for First Nations
- nomadic land tenure in Chad
- indigenous and gender-related land policy issues in Brazil
Ocean Governance: Use and Value of Marine Boundary Information
Sue Nichols, Michael Sutherland, Sara Cockburn, Sam Ng'ang'a, Ted Byrne, Meredith Hutchison, Karen Cove, John Hughes Clarke, Dave Monahan, Dave Coleman, Anne LaForest, Larry Felt, Gilles Paquet, Wade Locke and Ted McDorman
Examination of the role that property rights and marine boundaries play in ocean governance. Various projects including assessment of boundary requirements, the requirements for property rights and boundaries in ocean governance, property rights infrastructure in the offshore, and development of a database architecture for marine boundaries. Case studies included: the administrative boundaries of New Brunswick, including a legal background analysis; Canada's extended continental shelf; the boundaries of the proposed Musquash Marine Protected Area; offshore oil and gas leases; and aquaculture. The research also included numerous seminars, workshops, and presentations at the national and international level.
[GEOIDE, NRCan, SSHRC, CHS, Service New Brunswick, DFO, NSERC, ACLS]
Requirements for a Marine Geospatial Data Infrastructure
Sue Nichols, Michael Sutherland, and Katie Komjathy
To support ocean and coastal governance there is a need to improve the management of and access to marine information. This research will examine current information management activities in Canada, assess the needs related to developing a comprehensive management strategy, and recommend approaches to putting this infrastructure in place. It will be aligned with initiatives related to Canada's Ocean Action Plan and the Ocean Science and Technology Partnership.
[NSERC]
Adaptation Strategies for Potential Sea Level Rise and Storm Surges in NE New Brunswick
Sue Nichols, Michael Sutherland, Hazel Onsrud, Omer Chouinard, and Gilles Martin
This was a three-year project as part of an Environment Canada led examination of the potential physical, environmental, socio-economic impacts of sea level rise on the coast, including LIDAR mapping of the coastal zone. The adaptation component developed a strategy for helping coastal communities to identify appropriate adaptation options.
[NRCan (CCAF), Environment Canada, Parks Canada]
Linking Science and Local Knowledge with Coastal Communities
Sue Nichols and the UNB Land and Coastal Studies Group. Part of the Simon Fraser University Coastal Studies node of the Ocean Management Research Network
This research consisted of over 5 workshops on integrating scientific knowledge with community knowledge with Atlantic and Pacific coastal communities. Issues addressed included community capacity building, mapping, community resilience, sea level rise and storm surges, tourism, coastal planning, and First Nations.
[SSHRC and DFO]
Land Administration Reform in Chile and Peru
Sue Nichols with Watermark Industries and CPS
A World Bank stakeholder seminar was conducted and a position paper on strategies for improving land registration and administration in Peru was prepared. In Chile, a number of seminars and consultations were conducted with respect to national spatial data infrastructure and municipal GIS.
[World Bank, Industry Canada]
Development of a Marine Recreational Vessel Reconnaissance System
Kevin Pegler, Ph.D. Candidate; Marc Rancourt, M.Eng.; Katie Munroe, B.Sc.E. Candidate; David Coleman; Yun Zhang
Rescuing operators of small recreation vessels is a constant resource drain on the limited operating budget of the Canadian Coast Guard. As a result, a new and innovative application of small target surveillance techniques is being developed at the Department of Geodesy and Geomatics Engineering, UNB, Canada. This work is being done in support of the development of a strategic decision-making tool based on risk modelling to be used to predict where in Canadian waters marine incidents are most likely to occur in support of best resource allocation.
Previous research in the use of hyperspectral imaging for search and rescue, resulted in the development of fast, non-parametric "spatio-spectral" template subpixel object detection algorithm. This work (by Pegler) is being adapted and enhanced for use with the new, commercially available spaceborne high-resolution optical imagery. Investigations are being made regarding the utility of the Minkowski distance metrics for use in small target detection within a multispectral imagery environment. Further, research is being performed on the development of a weighted Euclidean distance metric (which is an adaptation of the Mahalanobis distance metric) to enhance the "spatio-spectral" template by exploiting the variance/covariance information surrounding a potential target. Finally, related research has led to development of semi-automatic water-masking techniques (Munroe) and an IKONOS coverage prediction algorithm and software package (Rancourt).
Optimizing Web Mapping Systems and Property Databases for E-Governance
Teresa Tang, M.Sc.E. Candidate; Jianfeng Zhao, M.Sc.E. Candidate; David Coleman
Canadian taxpayers invest tens of millions of dollars annually on federal and provincial mapping and/or property-related database development and maintenance, and still more on related "E-government" initiatives across Canada. Despite this substantial investment, many such datasets and databases - redefined some 10-15 years ago primarily to make digital maps - are neither structured nor customized appropriately to meet the future demands of e-government and e-governance.
The objective of this research program is to develop and evaluate a collaborative, Web-based architecture and system prototype which will: (1) more effectively employ mapping and property-related databases electronic delivery of government services; and (2) enable citizens to provide more in-depth feedback to government through the use of enhanced, easy-to-use Web-based mapping and analysis tools (Tang). Also, subsequent investigations of the applicability of social networking tools and technologies will examine the tradeoffs between viewing this problem from "people-driven" versus "issues-driven" perspectives (Zhao).
Developing such a system will enhance the role of key government mapping databases and - ultimately - help provide the infrastructure for more timely, direct, and informed citizen engagement in land management policy, planning and decision-making.
Algorithm and Software Development for Precise Road Change Detection by Integrating Ikonos and QuickBird Panchromatic and Multispectral Images
R. Wang and Y. Zhang
This research explored an effective and low cost way for precise detection of road changes using panchromatic and multispectral high-resolution satellite images, such as Ikonos and QuickBird. The research was concentrated on the integration of panchromatic and multispectral images for precise detection of road changes.
[GEOIDE]
Automatic Colour 3D Satellite Image Generation and Dynamic Colour 3D Online Visualization
P. Xie, H. Li and Y. Zhang
In April 2005, Google Maps first published 2D satellite images through its online mapping technology, resulted in worldwide breaking news such as BBC News (http://news.bbc.co.uk/2/hi/technology/4448807.stm) etc. In January 2005, a research project funded by NSERC I2I was started in GGE at UNB led by Dr. Y. Zhang to develop a breakthrough technology for automatic generation of colour 3D satellite images and dynamic online visualization. In December 2005, a Website for dynamic, multi-scale colour 3D satellite image visualization and online mapping was successfully launched (http://studio2.gge.unb.ca/3DSatelliteImage/). The Website shows 2D and 3D colour satellite images of the Fredericton and the region at different scales, from individual 3D trees, 3D buildings to 3D valley of Saint John River. The technology can be used to generate 3D satellite images covering the globe for worldwide online 3D visualization.
[NSERC I2I]
New Technology Development for Automatic Generation and Optimization of Web Colour 3D Images
P. Xie and Y. Zhang
In this research, a Web-based 3D system prototype was developed to allow users across the world to access and visualize colour stereo images of an interested area. Three modules were developed to complete this system: (1) Automatic epipolar image generation; (2) Automatic optimization of stereo viewing; and (3) Dynamic transfer and interactive visualization of the colour stereo images on the Web.
[NSERC]
Moving Target Detection Using Single Sets of High Resolution Satellite Imagery
Z. Xiong and Y. Zhang
This project aims at the development of a new technology for detecting moving targets and their speed using a single set of high resolution satellite imagery such as Ikonos and QuickBird. The resulting technology will be a semi-automatic or automatic system for moving target detection which can detect moving targets and their positions, moving directions and speeds with respect to a specified coordinate system. Precise satellite sensor models have been developed to accurately calculate the position and speed of individual targets. Initial manual testing results have proved that the concept of this project is promising.
[NSERC]
Incorporating the Challenges of Multiple Representation and Generalization Into Formal Description of Production Workflow
Xiaolun Yi, M.Sc.E. Candidate; Yvan Bédard (Laval University); David Coleman
New Multiple Representation Database (MRDB) systems are beginning to allow multiple geometric representations of the same objects to coexist in one database. How to build a MRDB is a challenge in itself. Traditionally, this has been achieved by creating different spatial databases and capturing each geometric representation separately. With database-driven approaches to digital map and GIS data production now becoming available, it is possible to capture these representations of specified features at the time of original collection in a manner that saves significant time and cost when compares with more mainstream digital map production methods.
Building a MRDB may involve capturing a new dataset from scratch, and formally describing the production workflow involved in such a procedure is a challenge in itself. Currently there is no existing software to support multiple representation data capture.
This research develops a conceptual workflow of multiple representations data capture which involves three phases: modeling, measurement and integration. It gives a dynamic model of multiple representations for data structure that incorporate the important concepts of explicit object and implicit object. In addition to formally describing a Multiple Representations Data Capture (MRDC) workflow using Unified Modeling Language, the project also includes development of a prototype system of multiple representation data capture based on this workflow.
Technology Development for Imaging and Generating Combined Colour 2D and 3D Images
Y. Zhang
A novel technology for imaging and generating simultaneous two-dimensional (2-D) and three-dimensional (3-D) colour remote sensing images was developed. The image can be seen in normal 2-D colour but also in 3-D colour when inexpensive stereo glasses are used. Unlike conventional colour 3D visualization techniques, the new technology allows the colour 3D information being display on a screen and also printed on a paper. Data volume of the new 2D and 3D image is significant less the conventional colour stereo images and no digital elevation models are required for the colour 3D visualization. A US patent application was filed in October 2002.
Automatic Fusion of Satellite Panchromatic and Multispectral Images
Y. Zhang
A breakthrough automatic fusion technique has been developed by Dr. Zhang that solved major problems existing in the existing fusion techniques. The panchromatic and multispectral images from Ikonos, Landsat 7, QuickBird and other satellites can be automatically fused, resulting in optimum fusion results with:
- almost identical colour to that of the original multispectral image;
- the same spatial detail of the panchromatic image; and
- naturally (smoothly) integrated colour and spatial detail.
Patent search results demonstrate that this technique is novel. The industry giants Kodak, the developer of the new satellite sensors for Ikonos and QuickBird, and Space Imaging, the owner of the Ikonos satellite, hold two patents for fusing images of their sensors, respectively. But, the new fusion technology developed at GGE is generic and is able to automatically fuse images from all optical sensors. A US patent application was filed in January 2003.
The image fusion technology has been licensed the world leading image processing and remote sensing software company PCI Geomatics. "We are very impressed by the new pan-sharpening technology, which is clearly leading edge and very exciting," states David Stanley, Vice-President of Research and Development for PCI Geomatics (http://www.pcigeomatics.com/pressnews/2003pansharp.html).
UNB also signed an agreement permitting DigitalGlobe to incorporate the algorithm into its production line to produce pan-sharpened QuickBird images, which are currently commercially available colour satellite images with the highest resolution in the world. Google Maps uses the pan-sharpened QuickBird images to show details of selected areas and cities.
[GEOIDE]
Development of Techniques for Colour Enhancement of Satellite Images
Y. Zhang
The "natural" colour composites of most natural resources satellites do not reproduce true natural colours as found in the nature or on a colour photo. In this research a novel way has been discovered to enhance the natural colour of satellite images for a better visualization effect. The concept of this method has been proved by processing original multispectral satellite image sets, resulting in significantly better visual effect.
Interactive and Dynamical Web Image Maps for the Public of Fredericton
Y. Zhang and P. Xie
In this project the latest Web technology and ViewPoint technology were integrated to develop innovative Web sites that allow users to interactively and dynamically explore a 1m colour satellite image map covering the urban area of Fredericton. The latest automatic image fusion technology developed Dr. Zhang was employed to generate the 1m Ikonos colour image with a highest quality. The Web sites are developed to serve for the public and schools in Fredericton with a new means. A tremendous member of positive feedbacks has been received from the public and schools. The Daily Gleaner newspaper reported the research results on February 1, 2003.
[UNB]