
HYPERSPECTRUM NEWS LETTER
Volume 1, Number 1
The above AVIRIS data-cube was taken from NASA/JPL FTP site
TABLE OF CONTENTS
What do scud missile launchers, cancer cells, crops, geological structures, camouflaged targets, organic pollutants, illicit drug manufacturing byproducts, rocket plumes, airborne targets, counter countermeasures, and chemical warfare agents have in common? Read on if you really want to know.
A new industry has spawned in the past few years, fast spreading into multidisciplinary applications. The growth is driven by a sophisticated market asking "what can it do for me?" and by technological developments in sensors, computers, data recording, signal processing tools, and more. An outgrowth of multi-spectral imaging, a cross between imaging and spectroscopy, the field is known as imaging spectroscopy (IS), or hyperspectral imaging. (This falls under the category of "What do you get when you cross a snake and a porcupine?" Barbed wire! You see what I mean about imaging spectroscopy?)
Some of the disciplines utilizing IS are listed in the logo above. Although the phenomenology is quite different, striking similarities can be found in the sensor concepts, data collection and recording needs, signal preprocessing, analyses tools including statistical and deterministic techniques, chemometrics, etc. (a more detailed comparison in a future issue). This newsletter may be the only tool dedicated to these common issues, promoting cross pollination and dissemination of information among interested parties.
We're interested neither in spatial imaging nor in spectroscopy, but in techniques that combine knowledge in both fields. So what? Why a newsletter? Here are a few possible answers; choose the one or two that mostly fit your situation. If you do not have affinity to a particular one, or can't supply your own answer, then obviously this newsletter is not for you.
- The technologist type -- I want to know if background clutter rejection techniques used in spectral signature analysis of cancer cells can help me find scud launchers in the desert (or vice versa)
- The curious type -- just want to know what is going on
- The manager type -- so many organizations involved, I can't keep track of everything anymore (information overload)
- The skeptic -- I think we may be duplicating work done by others (re-inventing the wheel)
- The marketeer type -- I have a piece of the puzzle and am looking for potential users
- The entrepreneur type -- just looking for new applications for my present capabilities
- The rest of the crowd -- current industry's news do not have a home and are scattered through various magazines or conferences.
Nahum Gat, Editor
How do I know? I'm just trying to facilitate the communications. It is you, the interested readers, the practitioner in the field, who should set the tone and context to the newsletter. These days, multispectral systems and applications are like a multicultural society. There is a little of everything in it, and it is PC (politically correct). So, expect anything. Here are some suggestions, however.
- Industry news, including new projects, applications, contracts, people news
- Brief technical write-ups discussing a concept, specific technology, approach, design, application, etc. (remember, though, we really do not want to take the business away from the glossy scientific journals and trade magazines)
- Data formats for exchange of "data cubes"
- Sources of information and test data
- Have a tough technical question? Maybe our readers have an answer
- Upcoming events of interest.
- The figure on the right is my concept of Imaging Spectroscopy.
WHO, AND WHAT DISCIPLINES ARE (OR AREN'T) COVERED BY IS
A corollary to this question is "When does multi-spectral turn into hyper-spectral?" There isn't enough space in this publication to provide an authoritative response to this deep philosophical quandary. But in general if you deal in electro-optical systems, and claim a portion of the spectrum between the ultraviolet to long wave infrared in more than X bands, and also deal with imagery, it'll buy you a ticket to the club. But if your personal X<2, well, just keep it to yourself (2 is really generous). A graphical orientation of the instrumental technique is shown in the figure on this page.
Oh yes, one more important constraint. Your brain must be programmed to operate in wavelength. If yours operates in wavenumbers, you're just a spectroscopist in disguise. I'm not aware of any OTC cure for that, but after all this is not like being born righty or lefty. On a second thought, however, it is.
Other than that here are a few areas that use (or could use) hyperspectral imaging.
- Space- or air-borne remote sensing of land and oceans
- Remote sensing of pollutants, and other compounds in the atmosphere, soil, and waters
- Medical photo-diagnosis, also known as optical biopsy
- Military target detection, recognition, etc.
Any omissions in the above list are deliberate but may be reconsidered in the future.
This is the tricky question. You can fill in the information in future editions (consider this an opportunity to promote your viewpoint at no cost, other than your creative time).
But we would like to hear from you if you do any of the following:
- Build imaging spectrometer sensors (staring or scanning)
- Use gratings, prisms, tunable filters, or other dispersive elements in your sensors
- Develop detectors and focal plane arrays applicable for imaging spectroscopy (IS)
- Design and/or build optical systems, scanners, etc. for IS
- Operate complete sensors
- Provide data or other services
- Develop software for IS
- Develop algorithms for signal processing and data utilization
- Have interesting applications
- Develop phenomenology models
- Develop hardware for data acquisition and recording
- Use interesting computational resources
- Interested in passive or active systems.
An across-disciplines newsletter may benefit us all in unexpected ways. In short what I've been trying to say, is that if you are in this business you can't effort not reading this newsletter, and that you certainly want to have your name in it.
To rephrase the question faced by Alice upon running into the pipe-smoking caterpillar... Your loyal editor, Nahum Gat, is coming to you from the white marbled board rooms of the industrial conglomerate known as Opto-Knowledge Systems, Inc., aka OKSI. Located in mid-town Manhattan Beach, CA, a short jog away from the white sandy beach, home of the annual world cup Beach Volleyball Championship, and within an imaging distance of the snow covered peaks of Southern California (that is if you select the right wavelength to penetrate the local atmosphere that contains a blend of common O2/N2 plus a garden variety of compounds that will guarantee a job security to an aspiring remote sensing spectroscopist).
HOW TO COMMUNICATE?
Questions, ideas, inputs, short write-ups, etc., can be communicated via phone, fax, snail-mail, or e-mail (a highly preferred approach -- I have nothing against ATT making money on long distance, but those bits and bytes use toll-free routes). We may even consider setting our own WWW node for you, Mosaic browsers (get the post office out of business). To save us the retyping of your inputs (and adding to your own typos), the preferred mode of communications for newsletter inputs is e-mail or a 3-1/2" diskette. Graphics, art, and photographs are acceptable in standard computer file formats (tiff, gif, etc.).
Contact: Nahum Gat, at the address, phone, fax, or e-mail listed on the cover.
And one last word; if you're concerned that your input will not make it into this space- limited, prestigious, archival, publication due to the overwhelming response... don't worry. Our editorial staff will save no effort to assure that all possible materials are encapsulated as eternal records to the pioneering days of imaging spectroscopy. The strict peer-review will be fair and swift.
Hyperspectral Analysis
A concept that always attracts the listeners' interest is the following definition (taken after Elachi, C., Introduction to the Physics and Techniques of Remote Sensing, Wiley Intersciences, 1987). A high resolution signature in the spectral domain can be transformed into a "hyperspectral" domain using various mathematical transforms. This allows treatment akin to the familiar time-spectrum domain analysis. Such hyperspectral transformations may provide the algorithmic basis for signature detection and recognition. More on these techniques in a future issue.
Dear The Expert: As I increase my spectral resolution, my signal level goes to pot. I'm so worried I can't fall asleep. Any ideas?
Dear Hyper: yes, count photons.
Join our staff of reporters and tipsters around the country. Send in any news worthy of printing.
If this issue is not addressed personally to you, and you wish to receive your own copy, or need an address correction, please e-mail or fax us a note.
INAUGURAL ISSUE , DECEMBER 1994
Competition of the month: Tell us how your favored algorithm picks up Lola under this poor target/background contrast (Harold is the decoy). TBD
OPTO-KNOWLEDGE SYSTEMS, INC.
1737 Third Street
Manhattan Beach, California 90266-6308
310-372-6665 FAX# 310-379-9842 E-mail: oksi@cerfnet.com
Copyright: © You MAY copy this newsletter to your computer or print it for your own entertainment or amusement, or even for business and technical purposes. You MAY NOT mirror or redistribute the newsletter on another WWW site. You can, however, include a link to the newsletter on this site in your own documents.