Mladen Stojic, President of Hexagon Geospatial
Dynamic Panoramic is a column by Mladen Stojic, President of Hexagon Geospatial. Mr. Stojic is a renowned geospatial industry expert who shares his insights into how geospatial information is increasingly being used to drive critical decisions in organizations of all sizes.
Better Urban Change Detection
January 16th, 2014 by Mladen Stojic, President of Hexagon Geospatial
Geospatial questions abound in urban areas. Property Assessors need to know who built an extension to their home to apply the appropriate tax and keep the city coffers full. Urban Planners are interested in green space, changes in green space and whether it provides the necessary continuity for birds and other urban wildlife. Utility companies often look for new construction, and indicators of affluence such as swimming pools to determine the future energy demands of a particular neighborhood.
They may also use LiDAR to manage vegetation encroachment on utility lines. The Water Department needs to understand runoff from impervious surfaces and how this impacts the underground infrastructure of pipes and drains. The health of water bodies is important for health and safety reasons. In some countries, sunlight is highly valued and any construction that casts a shadow on the neighbors is illegal. Transport departments are interested in the road conditions, closures and alternate routes. Emergency services need to understand various flood or storm surge scenarios, in some neighborhoods the fire potential, or simply an understanding of where the crimes or accidents are prevalent.
Intergraph Geospatial 2014 provides powerful geospatial analytics to the domain expert, the utility worker, the tax assessor, the urban planner. The tools are task oriented, connected to a database, portable and always reliable, accurate and based on sound technology.
Sometimes the geospatial query is complex and requires a geospatial expert to create models. ERDAS IMAGINE’s spatial modeler enables graphical dataflow construction, rather than relying on proprietary programming and scripting languages. The IMAGINE Spatial Modeler enables you to create standardized models for file sharing through web portals.
The expert can model geospatial queries in real-time using a wide assortment of raster, point cloud and GIS operators. The utility of this tool is extended through Python, or the new generic command line operator. The geospatial expert can also expose his knowledge to the people in the field with a “build it once and reuse” philosophy.
All models can be exposed through a simple GUI where the end-user simply provides inputs, outputs and a few variables to expose. Even these can have smart defaults.
The expert can also publish the model as a Web Service. The model itself and all the inputs can be described so the user accessing the model online understands exactly what it does and how it is used. In addition, the expert can pre-define the data input requirements of the model so that on execution, the user is only presented with appropriate inputs. The model is then executed on a server and the solution is made available to the client application.
One example application for better urban change detection is described below:
Performing automated “change detection” in urban areas has always been a challenge. Advances in data acquisitions, algorithms and processing environments have led to better and more reliable results that can be used by non-specialists.
Start with a high-resolution digital camera capable of recording 4-band CIR data (R,G,B,NIR) and perform a photogrammetric technique called Semi-Global Matching. This enables you to derive very dense point clouds from stereo overlapping images to produce ortho-rectified data consisting of 5 bands of information. These five bands are the standard R, G, B, NIR, plus the SGM-derived elevation points. These elevation points are what is used to ortho-correct the EO imagery. This ensures that the height is precisely aligned with the image “feature,” eliminating one of the common problems associated with trying to use separate image and elevation sources. Then the image bands are converted to reflectance values so different image dates can be easily compared. This data product is then used in ERDAS IMAGINE’s spatial modeler to automatically detect changes in buildings and trees. The model can be published as a WPS (web processing service) in ERDAS APOLLO, enabling many people to access the solution.