SO, WE'VE COME TO THE END OF THIS TRAINING SESSION. And, almost to the end
of the Tutorial. But there is more that you can do. Included in the PIT-image folder is the same complete 7 TM band set
for Morro Bay, California, designated by "morobay#" (where # is the band number) that you encountered in Section 1 as an introduction
to Landsat imagery and to image processing. The full 7 band
set covering the Waterpocket Fold in Utah is labeled "rocks#". Review Sections 1 and 2 to
remind yourself of the image content and the classes established for each scene.
Then, treat them as unknowns and do your own classification (use the same classes
or define new ones) and any other modifications of interest to you. See how well the different PIT classifiers perform compared
to the IDRISI classifiers. Be adventuresome and run unsupervised as well as
supervised versions. Another image which you examined in the Final Exam (Section 21) is that of Bloomsburg, PA. It is identified by this: "Bmoom1~1.img", "Bmoom2~1.img", etc. There is also a Landsat TM image set covering part of New Orleans. It is labeled Band, but TM bands 1 and 5 are missing; still, there is enough useful data to conduct the usual processing operations.
So, what do we suggest you do with PIT and its images? As a minimum, these: 1) display individual bands and experiment with modifying contrasts; 2) make both standard and exotic color composites; 3) make several individual band ratio images (probably about 5) and then produce color composites from combinations of any three (good idea: change contrasts to optimize the mix); 4) produce PCA images and use the first 4 or 5 to make color combinations of any 3; 5) perform unsupervised classifications with or without TM Band 6; and 6) generate one to three supervised classifications, varying the number of classes sought, and either using class categories that were presented where the several scenes (MorroBay; Bloomsburg; Waterpocket Fold [rocks]) have been examined elsewhere in the Tutorial or choose your own class names (adventurous alternative). As you by now should have surmised, PIT is an excellent and easy to use image processing program well suited to training purposes. But it has its limits - for example it has no spatial filtering routines. Other programs can be downloaded from the Internet (thus, in the public domain) and still others can be purchased. The writer has investigated two available off the Web: 1) MultiSpec, developed by Purdue University's LARS facility, which can be downloaded following the screen instructions and 2) Eduspace's Home Page, which displays an Image Processing buttom and lets you choose LEOWorks as a free download; Eduspace is sponsored by the European Space Agency. Both programs have additional features compared with PIT; both will also handle hyperspectral images. Both have rather extensive training manuals that can be off-loaded. But, the writer has encountered some problems in using each processing program, owing in part to some confusing guidance given in the training manuals. However, if you are adventurous, try each out and save if they prove useful to your purposes. To get a feeling for what a major commercial image processing system can do for you, check out the ERDAS product.
