Accurate data creation and 3D object extraction for GIS databases from digital...

Accurate data creation and 3D object extraction for GIS databases from digital imagery

SHARE

Kaushik Chakraborty
ERDAS Inc. Telford House, Fulbourn
Cambridge – CB15HB , United Kingdom
Tel: +44-1223-884510, Fax: +44-1223-880160
[email protected]

Abstract:
The correct and regular assessment of the status of the urban environment is a fundamental towards the understanding of the transformations of local, regional and global ecosystems. Extension of built-up areas and the loss of green space both within and around urban areas threaten bio-diversity as well as the quality of life of the residents of urban area. In order to aid urban planners make accurate predictions and plans, it is imperative to have an accurate and sustainable spatial database. This database should be accessible to the planners via a Geographic Information System (GIS). Imagery of the Earth’s surface, both aerial and satellite, has become an integral part of today’s GIS and desktop mapping systems. While a traditional Vector Map is a representation of the landscape as defined by a cartographer, an image map shows information from the ‘Real World’. Imagery is no more just a pretty backdrop but a crucial information extraction tool in the Geo-Spatial world. Image-based information products are valuable tools for urban planning. The true usefulness and usability of a GIS stems from the accuracy of the data. Most of the existing maps, topographic sheets and therefore GIS databases existing today are becoming rapidly obsolete with due to rapid changes and inaccuracies. Traditionally digital photogrammetry has been used by mapping firms to create the base maps to update the GIS databases. These systems have been for non-GIS experts and such systems have cost more than what most GIS projects could afford. ERDAS has released desktop digital photogrammetry solutions (IMAGINE OrthoBASE and Stereo Analyst) which will aid urban planners maintain the base maps in urban geo-spatial databases accurately and cost effectively.

Introduction
A large population of India lives in cities and towns, and urban areas are therefore the places where the environmental problems touch most the quality of life of citizens. Indian cities are experiencing an increasing use of energy, natural resources, and living space. The demand for residential areas, social facilities, infrastructures , health facilities put pressure on available living space of cities. Today’s urban planners have to understand the complex of relationship between people, resources, environment and development. Given these circumstances, many planning agencies throughout the world using complex Geographic Information System (GIS) for data capture, storage and analysis for deriving more realistic models for development. GIS is a tool for integrating large volumes of spatial and other data sets on different themes for such analysis, and offers excellent tools for understanding urban development in the past and running prediction models for the future. Collecting geographic data is vital to creating and maintaining a GIS, because inaccurate or outdated information will not reflect true, real-world scenarios. Today’s GISs are built using information derived from various types of geographic data, chiefly vector-based data. The increasing number of new data sources requires new tools that can bring into clear view today’s complex world of 3-D objects, features and spatial interactions, and help build vital infrastructure for the future. However, data currency and accuracy poses challenges to all GIS professionals. How can the GIS be updated and corrected?

  • The costs and time required to prepare and collect GIS data from existing sources of information can be high. For example, geo-rectifying 500 photographs to map an entire county may take up to three months before data collection begins.
  • Digitising hard-copy maps is time- and labour-intensive, and it introduces errors into the system.
  • Most of the original sources of information in a GIS provide only 2-D information (the X and Y coordinates).
  • Outsourcing core digital mapping and routine updates to specialty shops is expensive and time consuming.

Fortunately geographic imagery provides the ideal solution for deriving current and accurate 3-D information, and has proven to be a cost-effective tool for updating a GIS. Digital photogrammetry is an essential imaging tool for creating useful 3-D data and building a 3-D GIS. The idea of integrating digital photogrammetry and GIS has intimidated many within the GIS community. The accuracy of photogrammetry is well known, but the cost and learning curve associated with it have forced many GIS managers to select less accurate methods in their production. Many local governments have resorted to outsourcing their projects to specialty photogrammetric production shops. Fortunately, advancements in the development of geo-spatial tools at ERDAS, Inc. are bridging this long-standing gap. It’s now possible to precisely identify a geographic location in 3-D space and link that location with its information attributes through the synthesis of photogrammetry, remote sensing, GIS and 3-D visualization. This combination of geographic imaging techniques, fuelled by the precision and accuracy of photogrammetry, are the ideal tools for building the 3-D GIS of the future.

From Imagery to a 3-D GIS
Transforming imagery into accurate 3-D data involves several processes. The data and information required for building and maintaining a 3-D GIS includes orthorectified imagery, digital terrain models (DTMs), 3-D features (i.e., vectors) and the non-spatial attribute information associated with the 3-D features. Using various processing steps, the 3-D data can be automatically extracted and collected from imagery. Digital photogrammetric techniques offer unlimited access to the various types of photography and imagery that can be used to collect accurate GIS data. Although traditional photogrammetry applications use aerial photography (commonly nine by nine inches in size), new photogrammetry tools from ERDAS, Inc. easily handle any type of imagery, including satellite, digital camera, video camera, and 35-millimeter camera photography. Hard-copy photographs must be scanned or digitised to the desired resolution. For highly accurate mapping projects, calibrated photogrammetric scanners will scan to high precision (i.e., microns). If high-end micron accuracy isn’t required, more affordable desktop scanners (i.e., A3 size to accommodate nine- by nine-inch photography) can be used. Conventional photogrammetric applications such as topographic mapping and contour line collection use aerial photography, but current photogrammetry applications extend the processing to include oblique and terrestrial (i.e., ground-based) photography and imagery. Various image file formats can be used, including TIF, JPEG, GIF, Raw and Generic Binary, and compressed imagery, along with various software-specific file formats. GIS Update Workflow for Urban Planners
Traditionally most photogrammetric / mapping houses users have updated GIS databases by first creating Orthophotos and then digitising 2D vector from those Orthophotos. ERDAS software can take the urban planner from raw imagery to 2D/3D GIS database without the need for creating Orthophotos or DTMs. The straightforward and linear process for this includes several steps:

  • Sensor model definition
  • Ground control point (GCP) measurement
  • Automated tie point collection
  • Block bundle adjustment (i.e., aerial triangulation)
  • 3-D feature collection and attribution

The workflow is scaleable and doesn’t need to be repeated for every application scenario. For example, a block bundle adjustment doesn’t need to be performed every time imagery is processed for 3-D feature collection. If the end user does need to create Orthophotos, this can be accomplished after the block bundle adjustment step. A sensor model describes the properties and characteristics associated with the camera or sensor used to capture an image at the time of capture. The internal (focal length, fiducial marks etc. for aerial photographs) and external characteristics associated with sensor geometry, the image and the ground are frequently known or can easily be determined. Increasingly, airborne Global Positioning System (GPS) and inertial navigation system (INS) information also is obtained during image acquisition. If these data are available, the external sensor model information can be directly input for use in subsequent photogrammetric processing. The user can then automatically compute Tie Points and proceed to calculate Orthophotos or 3D feature collection. Otherwise, most photogrammetric systems can determine the exact position and orientation of each image in a project using the block bundle adjustment approach. Unlike traditional geo-rectification techniques, GCPs in digital photogrammetry have three coordinates: X, Y and Z. GCPs can be collected from existing vector files (i.e., a road intersection), orthorectified images, DTMs, and scanned topographic and cartographic maps. GCPs are used in conjunction with block bundle adjustment to establish a geometric relationship among the images in a project, the sensor model and the ground so accurate 3-D data can be collected from imagery. Photogrammetry requires a minimum amount of ground control, and the number of GCPs will vary from project to project. A tie point is a point with unknown ground coordinates, but the point is visually recognizable in the overlap area between multiple images. Tie points are used to create geometric harmony among the images in a project so they’re positioned correctly relative to one another. ERDAS software uses Automatic tie point collection to automatically identify and measure tie points across multiple images and strips of imagery. Once GCPs and tie points are collected, block bundle adjustment can begin. Block bundle adjustment (triangulation) is essential to determining the information required to create orthophotos, DTMs, digital stereo models (DSMs) and 3-D features. An output report can provide detailed statistical reports outlining the accuracy and precision of the derived data. Once the block is ready and solved, the end user can then proceed directly to create /update the 2D or 3D GIS database. If Orthophotos need to be produced and printed for other applications, then this can be accomplished at this stage as long as a DEM (Digital elevation model) is available. Orthorectification is the process of removing geometric errors inherent within photography and imagery. Using sensor model information generated by block bundle adjustment and a DTM, GIS users can remove errors associated with sensor orientation, topographic relief displacement, Earth curvature and other systematic errors. Measurements and geographic information collected from an orthorectified image represent the corresponding measurements as if they were taken on Earth’s surface. 3-D data and information can be collected from Digital Stereo Models (DSMs). Using the block file created in the triangulation step, two overlapping images of a DSM are automatically aligned, levelled and scaled to produce a 3-D stereo effect (with appropriate stereo viewing hardware). A DSM allows GIS users to interpret, collect and visualize 3-D geographic information from imagery and is used as the primary data source for collecting 3-D GIS data. Geographic imaging systems like ERDAS Stereo Analyst directly collect true, real-world 3-D geographic information from a DSM, using a 3-D floating mark or cursor, and don’t require an additional DTM (Digital Terrain Model). During data collection, the 3-D floating cursor commonly floats above, below or rests on Earth’s surface or the object of interest within the DSM. To ensure the accuracy of the data, the height of the floating mark is adjusted so it rests on the geographic feature being collected. ERDAS Stereo Analyst uses a automatic Terrain-following cursor to automatically place the 3-D floating cursor on the ground so a user doesn’t have to manually adjust the height of the cursor every time a feature is collected. For updating a GIS, existing vector layers are commonly superimposed on a DSM and then edited and re-shaped to their accurate real-world positions. Two-dimensional vector layers can be transformed into 3-D geographic information. During data collection, the spatial and non-spatial attribute information associated with a vector layer can be edited, and the attribute tables can be displayed with the DSM. Automated attribution techniques simultaneously populate a GIS during the collection of 3-D data. Additional qualitative and quantitative attribution information associated with a feature can be input during the collection process.

Why move to a 3-D GIS database for urban planning?
With the evolution of GIS software packages, Urban planners can now have access to 3-D data in visualisation, spatial modelling and analysis applications. Output products created by 3-D geographic imaging techniques include orthorectified imagery, DTMs, DSMs, 3-D features, and spatial and non-spatial attribute information associated with a geographic feature. Using these primary sources of geographic information, additional GIS data can be collected, updated and edited. DSMs created from high-resolution imagery are useful for the following:

  • Identifying and categorizing urban and rural land use and land cover. 3-D topographic information such as slope, vegetation type, soil characteristics, underlying geological information and infrastructure information can be collected as 3-D vectors.
  • Accurately interpreting and collecting data such as soil type, slope, soil suitability, soil moisture, soil texture, surface roughness, etc. DSMs are useful for determining the suitability of a given development (i.e., highways, building foundations, etc.).
  • Estimating population using 3-D high-resolution imagery to identify the number of dwelling units for various household types. Determining the height of buildings is important to generate an accurate estimate.
  • Deriving house size, lot size, building density, street width and condition, driveway presence/absence, vegetation quality and proximity to other land use types for housing quality studies.
  • Identifying and inventorying geographic information for site selection applications such as planning of transportation routes, sanitary landfills, power plants and transmission lines.

Each application requires accurate 3-D topographic representations, geological inventory, soils inventory, land use and vegetation inventory, etc. Analysing the extent of urban growth in change detection studies requires photography collected from various time periods. Land use and land cover information is categorized for each time period and subsequently compared to determine the extent and nature of land use/land cover change. Accurate 3-D building polygons can be derived via digital stereo models with ERDAS Stereo Analyst

Conclusions
The backbone of a GIS requires a strong underpinning of geographic control that makes pinpointing an exact location, along with its attributes, possible. A 3-D GIS powered by digital photogrammetry presents a new paradigm that offers greater accuracy and precision in data collection, and preserves the investment made in a GIS by any Urban Planning authority or local government. The methodology discussed in this paper is currently used by a number of local government planning agencies to manage the accuracy of their geo-spatial databases with excellent results. ERDAS photogrammetric solutions are designed for the GIS user with minimal or no photogrammetry training so they easy to learn and use and importantly offer a cost effective way for local governments to maintain healthy landscapes.

References

  • Stojic, Mladen, 2000, 3-D GIS: Unleash the power, Geo-Europe November 2000.