Digital Image Processing with Freeware

 William S. Lowe, ES 775, Spring, 2008

"Free software'' is a matter of liberty, not price. To understand the concept, you should think of "free speech'', not "free beer.''   -  Richard Stallman

Abstract:

The past two decades have witnessed the development of many open-source and freeware applications in GIS, photogrammetry, image processing, and cartography. In many instances, these have become rivals and replacements for the commercial products available. This essay presents the image processing capabilities of three open-source and completely free software "packages", GRASS (Geographic Resources Analysis Support System) and ILWIS (Integrated Land and Water Information System) and Multispec32. All these products had their beginnings in and are now maintained by publicly-funded institutions. Images constructed with these applications are shown, demonstrating the the similarity of freeware products to commercial products and their sufficiency for most pedagogical and research applications.

The Free Software Revolution:

The branch of economic theory known as "welfare economics" reveals that whenever a good or service is priced above its marginal cost of provision, a welfare loss occurs to society, because the item is unavailable to many who are willing to pay more than the cost of providing it, but less than the price of purchase
(Watkins). Computer software is an example of this kind of economic inefficiency . The marginal cost of provision nearly zero, however, intellectual property law guarantees that the price remains high enough to exclude many willing to pay much more than marginal cost. Fortunately, the last few decades have seen a partial remediation of this situation in the form of the Free Software Foundation and its associated General Public License. The publishers of commercial remote sensing and image processing software have, in the special case of academic licensure, made their products available to the students and faculties of recognised academic institutions for the purpose of training those who may later be in positions to influence institutional purchasing decisions, and so make their products available at nominal cost for students presenting the appropriate dispensation. For the students, this will get them through the course, but not leave them with something of perennial use value, in the sense of a textbook.

Free Data:

Sources of free or low-cost data are rare, but can be found on the internet.  The datasets from which the Cape Canaveral and Oklawaha basin scenes depicted here were produced from data downloaded from Landsat.orgAnother limited source of free data, not only Landsat data but also ASTER, Modis, and Quickbird is landcover.org.


Freeware:

MultiSpec is an image processing application developed by David Landgrebe and Larry Biehl of the Laboratory for Applications of Remote Sensing (LARS),  School of Electrical and Computer Engineering, at Purdue University, West Lafayette, Indiana.  It was developed under the auspices of NASA and with the support of NAGW-925, NAGW-3924 and NAGW5-3975.  The Multispec project had its genesis in one of the first multispectral image processing systems, LARSYS, developed in the 1960's.  The design objectives of the current version were the ability to run on readily available, inexpensive processing equipment; ease of learning and use, and the importation and exportation of data in text and various image file formats. (Landgrebe).  One of the strengths of MultiSpec 32 is that it operates directly on public domain metadata and interchange formats, such as GeoTIFF.  This capability promotes the inter-operabilty with other freeware and commercial  applications, so that applications lacking the image processing capabilities of MultiSpec32 can import preprocessed images for their own presentation advantage.  The images presented below were first isolated from a much larger 1991 Landsat TM row 16, path 40 scene, processed in MultiSpec32, saved to a .jpeg format, then reduced in size and (first image) enhanced in both brightness and contrast with Gimp 2, the freeware GNU Image Manipulation Program.


ksc 1,2,3
  Cape Canaveral Landsat 1991 Bands 1, 2, and 3
natural color composite image

Cape Canaveral Landsat 1991 Bands 3, 4, and 5
false color composite image
ksc 1,2,3,4Cape Canaveral Landsat 1991 Bands 1, 2, 3 and 4
false color composite image
ksc 1,NDVI,3
Cape Canaveral Landsat 1991 Bands 1, NDVI * 256, and 3
false color composite image. An NDVI ratio image is substituted for band 2.
ksc isocluster
Cape CanaveralLandsat 1991 Bands 1, 2, 3, 4, 5, 7 isocluster image (16 clusters).

.ksc isocluster
Cape Canaveral Landsat 1991 Bands 1, 2, 3,4,5,7 isocluster image classified (10 image classes).

As seen in the images above, the MultiSpec32 application is capable of composite imaging, ratio band generation and imaging, and unsupervised classification using iterative isoclustering and reclassification. Other features not represented here include supervised classification with training sites, geometric and radiometric correction, and change detection. A more complete list of its features can be found at at the Multispec web site.


ILWIS is the Integrated Land and Water Information System.  As the title suggests, it is oriented to needs of the environmental science user community, so it is superlative in its user interface, analysis capabilities, and  image presentation features.  The ILWIS project was initially developed by the
International Institute for Geo-Information Science and Earth Observation, and is now maintained by 52° North Initiative for Geospatial Open Source Software, a private organisation publishing under both the General Public Licence and a dual licensing model for commercial ventures incorporating its products.

The striking feature of ILWIS is its similarity in appearance, functionality, and ease of use to commercial software packages.  It has been used in university curricula as a complete image processing package for student use, as well as for environmental and earth research studies by public and private agencies.  Shown below is the presentation of the Cape Canaveral and Kennedy Space Center bands 3, 4, and 5 composite image from the data source previously cited.

ILWIS 3,4,5


Although ILWIS is not a cartographic tool, its image presentation features are sufficient to document the image and convey the relevant spatial
information to the viewer, without resort to an external presentation product.  To make such a presentation image, the user constructs what ILWIS terms a "layout", comprised of a background, a raster image or images, titles and other annotation, scale bar, and north arrow.  An example is given below.  The scene is from the path 16, row 40 Landsat 1991 imagery cited.


ORB 1,2,3
 


GRASS: The
Geographic Resources Analysis Support System is a raster and vector based open source GIS application originally developed in the early 1980's by the US Army Construction Engineering Research Laboratories (USA-CERL).  Grass development was later ceded to an international development team and published under the GNU license, with the sponsorship of the Open Source Geospatial Foundation. Grass is a full-featured GIS, image processing, and spatial analysis application, with the exception of a map composer feature such as that found in ILWIS.  This deficiency can be overcome by the use of either Quantum GIS (QGIS), for which a GRASS plug-in is available, or the Generic Mapping Tools (GMT). Both are available for Linux and other Unix operating systems.  An important feature of GRASS is its ability to be networked to support several users working on the same project, using shared datasets  and protected  individual workspaces.  Shown below is the GRASS graphical user interface.


grass interface


Shown below is a landover analysis from the Spearfish, South Dakota dataset which is commonly used in GRASS tutorials.  By reputation, GRASS is somewhat difficult to set up, but in the author's experience, it yields to practice when the tutorials using the Spearfish dataset are followed.  GRASS was designed for use with the Unix operating system.  The image below was captured from a display using Linux, a freeware operating system developed by Linus Torvalds and available in many distributions by download.


landcover



Other Freeware Spatial and Image Processing Applications:


IRFANVIEW is an image viewer and file converter for image files supporting a great many image formats, as well slideshow production, scan and print, and many other features.  Irfanview is free for non-commercial use. 

GDAL, the Geospatial Data Abstraction Library, is a translator library incorporated into many freeware GIS and image processing applications, including ILWIS and GRASS.  It is also available with a command line interface for stand-alone use.  GDAL is released under an open-source license by the
Open Source Geospatial Foundation.

SAGA, the System for Automated Geospatial Analyses, is richly featured and user-friendly GIS having a growing scientific user community, with 120 modules in its standard edition.  It runs under most operating systems and is licensed under GNU.


Conclusions:

In this survey, a few of the image processing capabilities of three freeware image processing systems have been presented with images produced from each.  These applications all have their strengths and weaknesses, as do commercial applications.  Their inter-operability makes it possible for each to serve as an extension of the others.  For example, a six band image can be produced in Multispec32 and imported into ILWIS for presentation with the customary map furnishings.  For those wishing to continue their image processing beyond the expiration of their commercial licenses, or for those wishing to produce imagery in a commercial venture without the usual high start-up cost, freeware is an attractive option.

References:

"Marginal Cost Pricing", Thayler Watkins, San Jose State University Economics Department - http://www.sjsu.edu/faculty/watkins/mcpricing.htm.

"Against Intellectual Monopoly", David Levine and Michele Boldrin, - http://levine.sscnet.ucla.edu/general/intellectual/againstfinal.htm.

"MultiSpec Tutorial", Larry Biehl, - http://cobweb.ecn.purdue.edu/~biehl/Wetlands/MultiSpec_Tutorial.pdf.
 
"ILWIS Reference Guide" - I
nternational Institute for Geo-Information Science and Earth Observation
, - ftp://ftp.itc.nl/pub/ilwis/pdf.

"GRASS 6 in a Nutshell", Markus Neteler, - http://mpa.itc.it/markus/osg05/neteler_grass6_nutshell2005.pdf.