INTERACTIVE IMAGE PROCESSING

INTERACTIVE IMAGE PROCESSING

Guest Writer: Jeff Love, PIT Developer

Introduction

For those of you venturesome enough to work through this Appendix, we move now to what promises to be the most interesting and potentially informative segment in this whole Tutorial. You will be given the opportunity in this Section to actually process and analyze raw or altered space imagery using a NASA-developed (by Goddard’s Code 935, the sponsor of the Tutorial) image processing program. PIT, which stands for Photo Interpretation Tool, was developed primarily to support 935’s Regional Applications Center program (before it was transferred to Stennis Space Center). PIT is designed as a fairly simple, somewhat limited processing software system whose main purpose is to produce thematic classifications of Landsat, AVHRR and GOES image data sets;it may eventually also handle other sensor systems such as SPOT, AVIRIS, and JERS that generate multispectral imagery. In principle, it could also process multichannel radar data, provided each channel is co-registered with the others. PIT is in part a training package, although its classifications are commonly fairly sophisticated, but it is not as elaborate as such commercial programs as ERDAS and Idrisi. But, like those, it “puts the ball in your court” in that you can elect how to display an image set to its best advantage, choose what processing routines to apply, and define the categories to be established as classes in an image set subjected to maximum likelihood and other types of classifiers. In effect, you are now about to practice as a bonafide remote sensing technician in that you will use a powerful processor to control the modifications and optimal appearance of enhanced images and, with some guidance, interpret the features you recognize within them and extrapolate these decisions to achieve your own classification of scene content.

One thing you will not find contained in this Appendix: the PIT program itself. It proves technically difficult to embed a separate software package directly within the Tutorial. Of necessity, therefore, we are adding the PIT processing software, along with the images to be worked on, as an independent program on the CD-ROM; you must find the yellow folders in the CD-ROM labelled Pit and PitImages and get their content into your C:drive and onto your desktop with Pit being accessed through an icon on the desktop. There are two sets of instructions on how to get PIT running on your computer depending on whether you are using the CD-ROM or the Internet. PIT currently should work under Windows 98 and Windows NT.

Background: Many of the basic principles underlying image processing have been interpolated or treated directly in Section 1 of this Tutorial which shows how a single scene covering Morro Bay, California is image-processed using a variety of standard routines. It should be profitable for you to return to that section for a refresher on the methods and principles that will pertain to your tour through image processing in this Appendix. A readable summary of image processing, in more detail, is found in the writer’s Landsat Tutorial Workbook, which, unfortunately, is out-of-print and hard to find (some university libaries may have a copy). In the last 30 years, many books that include chapters on image processing have been published. Among these are: Lillesand and Kieffer, Remote Sensing and Image Interpretation, 5th Edition, J. Wiley and Sons; and Avery and Berlin, Fundamentals of Remote Sensing and Airphoto Interpretation, 5th Ed., Macmillan Publ. Several books entirely devoted to this subject also can be consulted if you wish a formal survey of image processing. One we can recommend is J.R. Jensen, Introduction to Digital Image Processing, 2nd Ed., 1996, Prentice-Hall. For a rather quick synoptic overview of Processing fundamentals, available on the Web, go to this site developed by the CCRS staff and/or to this basic review prepared by the GIS Development Group.

Here, in this Appendix, we will begin now by revieing or introducing in the next four pages some information on the nature of the data that are gathered first by the remote sensor, brought to a place for initial processing, and then organized into a format suited to reading by a computer-based image processing program.


Primary Contact: Nicholas M. Short, Sr. email: nmshort@nationi.net

Jeff Love, PIT Developer (love@gst.com)

Collaborators: Code 935 NASA GSFC, GST, USAF Academy