ENVI 4.4 is the latest edition of this popular image processing software from ITT Visual Information Solutions and builds upon the previous versions of this product. It is a complete image-processing system that includes spectral tools, geometric correction, terrain analysis and radar analysis while also incorporating strong GIS functionality for working with both raster and vector data. This version incorporates significant additions and benefits over previous versions due to its ability to work with a new ENVI Feature Extraction Module, Atmospheric Tools (FLAASH) and Digital Elevation Module (DEM) extraction. A new interoperable National Imagery Transmission Format (NITF) imagery format for enhanced data fusion is also included.
Image Processing for Geospatial Users
ENVI 4.4 builds on a history of excellence in software design and function for processing airborne images useful for geospatial purposes. During this evolution, user needs have grown and expanded. Improved technology, application requirements and demands, have also contributed to growth and change.
Many professionals involved in image analysis and processing must integrate image intelligence with other geo-technologies such as geographic information systems (GIS), global positioning systems (GPS) data, laser point cloud data, surveying and spatial information acquired through alternate technologies and sources, more quickly. Therefore, a federated approach, which involves many technologies, data types and sources, are all contributing toward greater challenges for integrating this information. This ultimately results in higher value from the data and technology.
To meet these challenges users have been demanding all the strengths of image analysis algorithms and functionality, while at the same time asking for them to be delivered through more easy-to-use interfaces. For software companies, the task has to been to deliver on these demands and development of tools with increasing levels of automation and ease-of-use has been the result. ENVI 4.4 has refined and incorporates these tools, providing greater ease-of-use (see Fig. 1), while also including user feedback from geospatial professionals and image analysis specialists in terms work flow demands. In speaking with Richard Cooke, COO and President, ITT Visual Information Solutions I inquired about ease of use, to which he said, “We wanted this release to be focused primarily on efficiency and automation – making image processing and analysis workflows easier and faster. ENVI 4.4 really provides over-arching capabilities to make the latest image processing technology readily available to anyone. The new workflow efficiencies will not only please experienced users, but will allow any level of user to perform advanced techniques.”
This has resulted in a user interface for this product that not only delivers on providing a simpler user interface with automation, but one that is logical in its layout and functionality. ENVI can be considered as a collection of toolboxes or functional components including:
Collectively, these tools provide a wide array of functionality, which enables users to interact with other geospatial information professionals, and their data and tools – or hardware. ENVI 4.4 while focused and optimized to deliver imagery related functionality, is not solely reliant on image data in its enabling functions. It can also import and manage a very wide variety of satellite and airborne derived data, including LiDAR sensors; this product can interoperate with GIS, GPS and other data obtainable through Open Geospatial Consortium (OGC) standards. Consequently, ENVI 4.4 can be viewed as a geospatial integrated data management software with particularly unique and strong image processing and analysis capabilities.
ENVI 4.4 can work both spatially and spectrally with individual pixels in imagery. This enables, for example, features to be identified both by their natural spectral characteristics as well as by their structure, shape and physical characteristics. This fact contributes to greater automation within the product because features can be assessed and classified more accurately based upon attribute identification characteristics associated with them. It also means that the software has enhanced filtering mechanisms for classification and identification purposes.
A full range of imagery sources can be processed in the ENVI 4.4 environment including AVHRR, Landsat TM, ASTER, MODIS, QuickBird, IKONOS, Orbview-3, and ENVISAT data, as well as dozens of other data types. ENVI is also capable of processing many other multispectral and hyperspectral images, and data from advanced SAR systems such as the new TerraSAR-X data products and those from INTERMAP, including NEXTMap Europe (see Fig. 2).
Accordingly, ENVI 4.4 provides tools, which enable classification of hyper-spectral images, providing the following analysis capabilities:
Each of these functions is accessed through the Main Menu bar. Selecting the File Menu allows for the selection of files in vector or raster formats as well as for initiating the ENVI Zoom Viewer (see Fig. 3) which includes new visualisation tools in this version for vegetation processing as well as vector layering. Connections to data warehouses may be made from within the Zoom Viewer. Alternatively, image processing functions are also available, including RX Anomaly Detection, Pan Sharpening, Vegetation Suppression and ENVI Feature Extraction – the later being of most interest in ENVI 4.4.
I imported a number of different types of satellite derived imagery products. For example, using Shuttle Radar Topographic Mission (SRTM) data and using one of the many available color ramps to shade some elevation data from the northwest coast area of The Netherlands. It was quick and easy to use the 3D viewer included.
ENVI Feature Extraction Module and SPEAR Tools
In addition to the spectral analysis tools through the Spectral Menu item off the Main Menu bar, the feature extraction functionality within this software is new. It provides enhanced capabilities for extracting information from imagery using both spectral and spatial filters – a unique approach. Both panchromatic and multi-spectral images may be processed for feature extraction.
I queried Richard Cooke on this point and he responded, “I think the most exciting feature for imagery users will be support for the new ENVI Feature Extraction Module, which is being released along with ENVI 4.4. We’re really excited about this technology, and it’s not available in any other product. It allows users to isolate and remove specific features – like buildings, vehicles, or bodies of water – and then use them for GIS, or to perform additional analysis using ENVI.”
Feature Extraction is accomplished by loading an image, selecting Feature Extraction from the processing menu and working with the Feature Extraction Wizard. Scale level is initially set and the area to be processed is selected (see Fig. 4,5,6). My attempts using this new tool were favorable. I was able to delineate features readily within 20 minutes or so without much difficulty. Indeed, the software includes wizards for identifying roads and boats in water automatically. However, users will find that they can build their own libraries for feature extraction based on the kinds of objects they wish to extract. Over time this software becomes quite valuable as the libraries take on increasing value for adding automation into the workflows.
The SPEAR toolset is designed to provide a higher level of automation, again allowing for a high level of complex analysis to be conducted by everyone. These tools include automated processes for spatial and temporal change detection, pan-sharpening images, and terrain categorization. Boats, for example, can be delineated with these tools, and water depths can be calculated. Working with the ‘Lines of Communication’ wizard (see Fig. 7), the SPEAR process is like ‘stepping into’ the manual process. What one would normally do in terms of setting spectral filters and adjusting them, SPEAR tools do automatically.
Once completed, each parameter can be fined tuned or exported. The benefit of this approach is that a large number of regional libraries for spectral analysis of a localized nature could be created and stored quickly. Such libraries can be readily shared, again speeding up the workflow processes by invoking analysis through the SPEAR tools wizards.
Many users are finding that that their workflows are encompassing higher levels of 3D orientation as well as being able to move towards the growing and expanding integration of GIS and CAD based tools and data. While GIS based information is particularly useful for geo-processing spatial relationships, particularly in a layered context; CAD information is often more closely associated with building structures, bridges, roads and other objects that are obtained through infrastructure design, surveying and or aerial data collection methods such as LiDAR.
These sources of data are being integrated together to provide geo-referenced objects on 3D landscapes (see Fig. 8). ENVI 4.4 contributes to this integration of shared data sources via its capabilities for building 3D digital elevation models. To accomplish DEM extraction, a stereo pair of aerial or satellite-generated imagery is used. The stereo pair must contain rational polynomial coefficients (RPC) positioning from aerial photography or pushbroom sensors. The RPC are used to generate tie-points, which then enable accurate positioning between images for DEM extraction. These models can be quickly built from ASTER, CARTOSAT-1, IKONOS, OrbView-3, QuickBird, and SPOT image sources.
The procedure is straightforward in order:
Clearly the ground control points and their quality will have significant impact on the output. In many cases the accuracy of the ground control points will be collected with GPS. Consequently, it is important to obtain as accurate as possible GCP. Of each of these steps, the eighth step is the one that I had the most interesting time with. Terrain relief estimation must be selected as a step in the wizard.
Values are low, moderate and high, for flat, rolling and mountainous terrain, respectively. To properly input this step takes some finesse, since some people may not be familiar with the actual terrain they have images for. However, GPS measurements will be able to provide an indication of the terrain. Some people, especially experienced analysts will be able to look at a stereo pair and indicate whether or not the terrain is flat, rolling or mountainous, just by looking at.
The level of terrain detail matching will vary, however, high correlation will provide higher levels of relief determination. Having said that, a DEM terrain-editing tool is including to alter obvious errors and to make adjustments. The actual map project of the output is accomplished through the reprojection of the DEM at the epipolar stage. The highest accuracy DEM will be a reflection of including GCP that highly accurate and matching between images in the stereo pair as well as accuracy in using the 3D Stereo Pair Measurement Tool.
The data table for elevation measurements is ‘extracted’ from the images in a ‘ENVI Point Collection’ table that consists of an ArcView GIS ASCII data table. This means that a table of coordinates (x, y, z) is being generated. Because the table is in ASCII format, it can be used in any GIS capable of handling and displaying 3-D generated data points. As can be seen, the entire process allows the user to collect, manage and select all data points – it is open.
As ITT Visual Information Solutions notes – the ENVI Fast Line-of-sight Atmospheric Analysis of Spectral Hypercubes (FLAASH) Module allows you to accurately remove the obscuring effects of the atmosphere. FLAASH was developed by Spectral Sciences, Inc., a world leader in optical phenomenology research, in collaboration with U.S. Air Force Research Laboratory (AFRL) and Spectral Information Technology Application Center (SITAC) personnel. These tools are new with ENVI 4.4.
Atmospheric effects have always been a concern to imagery professionals, for a simple reason. They obscure images and reduce clarity. This is particularly the case in urban environments where aerosols, smoke and other pollutants are generated in the highest amounts. Software that is capable of reducing these impacts means higher clarity, sharper images and improved intelligence. ENVI 4.4 is designed to reduce these effects pixel-by-pixel – since they are reproduced by individual image pixel.
The atmospheric type is calibrated against individual sensor type (satellite image sensor). Input radiance files are calculated and output reflectance measurements are determined. These corrections are then applied to the image and adjustment is made. Water vapor percent, latitude and time of year are all calculated into the corrections. Aerosol visibility ratings are also calculated. Finally, a recalibration of reflectance from image objects is determined and the image is adjusted accordingly.
Keep in mind that FLAASH is an ENVI Add-On Module and is not necessary to obtain very good results with the standard ENVI 4.4 software package. Only those applications that demand extended levels of atmospheric functionality (namely those in areas with high levels of atmospheric impact) or where more detailed visibility is needed will make use of these benefits.
National Imagery Transmission Format (NITF) standard is a raster format defined by the NITF Standards Technical Board. The Joint Interoperability Test Command (JITC) certifies systems implementing the NITF format for compliance with the standard.
This is my first encounter with this format, one that is designed for defense agencies and the intelligence community. It is sometimes also called the – NITFS. The added letter standing for the word ‘standard.’ This format differs from others because it is designed for interoperability between systems. For example, a package of information is included the data file that relates to the image itself. Most geospatial users are familiar with this concept; NITF goes one step further and can include overlay graphics, symbols and labels. It also includes Support Data Extensions (SDE), which allow space for added functionality pertaining to the image. Essentially then, this format includes the ability to move not only metadata with an image, but also the details pertaining to how it can further be processed etc.
Who else uses this format you wonder? Well, it can work with products from a number of suppliers including BAE Systems, DAT/EM Systems, DigitalGlobe, ERDAS, GeoEye, OSSIM, PCI Geomatics and so on. The Benefit to ENVI 4.4 is that it puts this software, and all the functionality mentioned above into an interoperable position with all these other supplier’s products and the governments and intelligence agencies themselves – where NITF data images are being used.
This review has focused on the newer additions in functionality related to the latest edition – ENVI 4.4.
Keep in mind that a lot of details in the basic software package have been quickly reviewed to write on the latest editions, much of which we can write about in the days ahead.
The latest editions to ENVI 4.4 are exceptional. The Feature Extraction Module coupled to the Digital Elevation Module (DEM) Module add-ons extends the value of this software considerably. Feature extraction is enabled through a unique wizard interface that enables users to integrate both spatial and spectral decision making into the image analysis process. SPEAR tools aid in developing quick applications, supporting existing libraries of information for automated processing, thus leveraging staff availability, time and cost. ENVI 4.4 includes FLAASH tools, which enable the highest levels of digital processing in trying environments where atmospheric and airborne pollutants degrade clarity.
The coupling of DEM Module automation with Feature Module extraction automation should not be overlooked – collectively they provide a full service and application value for 3D landscape generation. The addition of GIS functionality couples that value into a 3D geo-processing environment, again further extending the value of the data for decision making. This software is extremely stable, it did not experience problems once and the processing speed appears to be quite fast.
The Line of Communication wizards need to have adjustable box sizes. This would avoid their taking up valuable screen area, although they can of course be minimized. The manuals for this software are included. There is merit in showing more pictures with descriptions so one can conceptualize what they might be looking for in some of the automated analysis. As an aside, I wonder when ENVI will include 3D objects for placement on this highly accurate landscape – essentially moving this program into visualization more fully.
ENVI 4.4 packs a lot of analysis power within an easy-to-use interface. The connection this software makes with GIS and other geospatial data types is not only through connection, but someone has thought through the workflows that an integrated imagery-GIS professional would use. Consequently there is higher value in this software than previous versions as more data can be harnessed from imagery to decision-making. The feature extraction abilities will cause users to open up their historical libraries for new data extraction possibilities. Meanwhile, ENVI 4.4 opens the door to realizing the full value potential of current and near future data gathering efforts and sources with it enhanced processing and automation capabilities. If you are considering new image analysis software, then ENVI 4.4 should be on your list to consider.
Manufacturer: ITT Visual Information Solutions (www.ittvis.com)
Version: ENVI 4.4 with Feature Extraction, FLAASH and DEM Modules
Manuals and Literature: Included digital and online (www.ittvis.com)
Seminars and Education: (www.ittvis.com/webinar/index.asp)
User Symposium: May 20, 2008 – Virginia, U.S.A.
Jeff Thurston is Europe, Middle East, Africa and Russia editor for V1 Magazine. Based in Berlin, Germany he has reviewed leading software and hardware in the geospatial industry for over 10 years.