HobbyKing.com New Products Flash Sale
Reply
Thread Tools
Old Jan 22, 2015, 02:52 PM
Registered User
Near Austin
Joined Dec 2001
5,449 Posts
Discussion
Cadaver Detection - NIR Photography

Something I have been talking about for awhile on here.

http://www.discovery.ca/Shows/Daily-Planet

Select the Jan. 21st episode. Segment starts at the 06:15 mark.
CenTexFlyer is offline Find More Posts by CenTexFlyer
Reply With Quote
Sign up now
to remove ads between posts
Old Jan 22, 2015, 06:26 PM
Registered User
benedict's Avatar
United States, HI, Waimea
Joined Feb 2013
617 Posts
They broke it up into segments. "Spectral Drone" is the one to watch.

Cool segment! And some serious good press for the work you're doing.

I used to live in Austin, and spent a fair bit of time at my brother-in-law's ranch a little further west of there. He'd occasionally lose a cow, but there was almost always something that had died in the last few years that was decomposing out there. Aside from overall body mass, have you found a good way to reject false positives in your imagery coming from cows, dogs, or other critters that might leave a body on the landscape? Or are they so sparse that from the standpoint of search and recovery, it's still worth checking them out?

Have you tried combining your NIR imagery with color visible imagery? Judging by the ground footage in that segment there seem to be some pretty strong characteristic colors at various stages of decay.

Sorry to pick your brain like this. I hadn't really thought about using NIR for cadaver detection, and I seem to have missed the threads in which you wrote about it. So this is cool stuff from where I'm sitting!

Thanks again for sharing.

Tom
benedict is offline Find More Posts by benedict
Reply With Quote
Old Jan 22, 2015, 11:02 PM
Registered User
Near Austin
Joined Dec 2001
5,449 Posts
Tom,

That's one of the many things we are trying to discover. False color imagery seems to be a pretty good route on top of the NIR photography. Lot's of different things you can do even in PhotoShop to bring out certain characteristics. We still just don't know how much we can tease out of a photograph in any wavelength but we're finding out!

And you are correct in assuming that even though there is a lot of stuff out there that keels over in the wild, when you find those "islands" they are pretty few and far in between and should be checked out. There is also the physical dimension of the affected area. A 1200 lb heifer leaves a different print/signature than a 150-180lb human.

Kind of grisly stuff, but one of the things that we have learned is that the faster you are on a scene, the more you can preserve and determine COD and maintain evidence for the investigation. Something a photo shooting "drone" can collect quickly - much more so than a manned unit just "looking" for someone.

Gene

P.S. the ground footage int he piece was post process "tinted" to appear in the red spectrum. It was not actually NIR video.
CenTexFlyer is offline Find More Posts by CenTexFlyer
Reply With Quote
Old Jan 23, 2015, 01:03 PM
Registered User
benedict's Avatar
United States, HI, Waimea
Joined Feb 2013
617 Posts
Most of the NIR and four-color work I've done has been for vegetation study or art photography. Can't say I've ever gone looking for cadavers. But there's an amazing amount of information to be had from getting all four channels at once. The only real impact on the aircraft is that you're doubling the weight of the camera payload.

I've had the luxury of processing photos when I get home, but I don't think there's anything in the workflow that would preclude doing the processing on a laptop on-site. I don't know if this is close enough to real-time for what you're doing, but if you're interested I can put something together to describe how I'm doing it.

Actually, I had one other question if you don't mind: This whole technique revolves around the vegetation islands created by decomposition. In areas where vegetation is sparse, or in a place like a dry wash, is the effect strong enough that airborne seeds will germinate in an otherwise barren area, or for dormant plants to become active? Or is that kind of like looking for footprints on dry basalt?

This is fantastic stuff, Gene. Thanks again for sharing. You've seriously piqued my curiosity.

Tom
benedict is offline Find More Posts by benedict
Reply With Quote
Old Yesterday, 12:39 AM
Registered User
Near Austin
Joined Dec 2001
5,449 Posts
That's why I posted! Curiosity brings the bright minds in on this problem!

Quote:
Originally Posted by benedict View Post
Can't say I've ever gone looking for cadavers.
Yes, that's not something that most people get involved in, but it seemed to be a natural progression for us in the use of UA in SAR/Recovery. And it is fascinating too!

Quote:
Originally Posted by benedict View Post
But there's an amazing amount of information to be had from getting all four channels at once. The only real impact on the aircraft is that you're doubling the weight of the camera payload.
Actually, Tetracam makes a camera that will give you all those channels and it weighs less than some POS cameras. It is only a 1.3mp cam though which is a serious drawback.

Quote:
Originally Posted by benedict View Post
I've had the luxury of processing photos when I get home, but I don't think there's anything in the workflow that would preclude doing the processing on a laptop on-site. I don't know if this is close enough to real-time for what you're doing, but if you're interested I can put something together to describe how I'm doing it.
Absolutely yes! I'm all about learning different ways to post process!

Quote:
Originally Posted by benedict View Post
Actually, I had one other question if you don't mind: This whole technique revolves around the vegetation islands created by decomposition. In areas where vegetation is sparse, or in a place like a dry wash, is the effect strong enough that airborne seeds will germinate in an otherwise barren area, or for dormant plants to become active? Or is that kind of like looking for footprints on dry basalt?
Well..... that's another whole different ball of wax. When you get into areas of sparse vegetation, you're not looking so much for changes in the plants as you are a different sort of compaction of the earth, which a single channel of NIR like our 660nm wavelength camera will pick up. But even at that, I think we can improve on that. We discovered some Ottoman empire grave sites and archaeologically significant structures in Mongolia using that.

Quote:
Originally Posted by benedict View Post
This is fantastic stuff, Gene. Thanks again for sharing. You've seriously piqued my curiosity.
Yeah, ain't it just!

Gene


Tom[/QUOTE]
CenTexFlyer is offline Find More Posts by CenTexFlyer
Reply With Quote
Old Yesterday, 06:36 AM
Registered User
Philippines, Calabarzon, San Pedro
Joined Jul 2012
4,561 Posts
The video you have selected is not available for viewing outside of Canada. Damnit.
Hajile is online now Find More Posts by Hajile
RCG Plus Member
Reply With Quote
Old Yesterday, 09:01 AM
Registered User
Near Austin
Joined Dec 2001
5,449 Posts
Bummer dude.......

Discovery won't give a copy or let me link to it either.
CenTexFlyer is offline Find More Posts by CenTexFlyer
Reply With Quote
Old Yesterday, 12:08 PM
Tesla's Workshop
workshop's Avatar
USA, CA, Torrance
Joined Nov 2004
2,460 Posts
Gene, I work at Monarch, Inc. now and we are a Tetracam dealer and we have on-staff agronomists that assist us in tailoring NIR sensors to look for very specific bands for AG use (Monarch does large scale AG). Please let me know if there is anything we can do to help.

Jeff Parisse
MonarchMakers.com
workshop is offline Find More Posts by workshop
Reply With Quote
Old Yesterday, 12:17 PM
Registered User
benedict's Avatar
United States, HI, Waimea
Joined Feb 2013
617 Posts
I'd never heard of Tetracam before this. Looks like they've got a couple, some of which go up to 3.2MP. That one aligned array looks great, but heavy. HOLYCRIPES! Ok, their pricing took me a little by surprise. Scratch the aligned array. I'll leave that for deeper pockets.

So far my multi-spectral camera array is about $60 worth of parts off Ebay and a Fiji script to align the RGB and NIR photo pairs. A lot less convenient than a self-contained camera, but a lot less stressful to put in the air.

As for processing, Fiji is a nice compromise between manual and semi-automated. You can process single photo pairs or batch process entire SD cards worth of photos. Fiji is basically a nicely packaged distribution of ImageJ that people have written macro packages and plugins for. It comes with a bunch of stuff pre-canned into the distribution, so if you're looking to install one or the other, go with Fiji.

One plugin I like to use is DStretch. It was written by Jon Harman to do analysis on pictographs, but I've used it for aerial analysis. DStretch lets you do principal component analysis on any digital images. The way PCA works is to map all the pixels in the image into three dimensional color space (R is one axis, G is another, and B is the last). In most images this color "shape" is something like an ellipsoid. It then performs a coordinate rotation on the color space to align the strongest difference in color to one of the primary color axes (e.g. put the long axis of the color ellipsoid to the R axis), the next strongest difference to another axis (e.g. put the larger of the two minor axes on the G axis), and the third falls where it will. Even better, it does all this automagically, so that's the last time you ever have to think about how PCA works. Suffice it to say it makes subtle differences in color pop out at you.

Because DStretch runs inside ImageJ, you can use it to process single images or you can use it to process entire cards full of images.

The last processing tool I tend to lean on is Imagemagick. It's a command line image processor that makes batch processing very very straightforward. It's my go-to tool of choice for doing weird channel math because it treats everything as a floating point number. So you don't get weird artifacts in your images because you run out of bits in the middle of a calculation. Of course its worst drawback is speed because it treats everything as a floating point number. But still, it's one of the few packages I've found that lets you do things like taking the red channel from one image, subtracting it from the green channel of another, and then dividing it by some other combination of channels, and mapping it into a particular color on your resulting image.

All of these are free to download and use. The biggest cost involved is the learning curve to get up to speed on them. Fiji is pretty straightforward. It's an image editor. DStretch is dead nuts simple once you get into it, but I spent a couple of days just playing 'cause it's fun to use. Imagemagick will be familiar(ish) ground if you spent a lot of time using command line stuff on UNIX or back in DOS days. For someone raised on a windowed operating system it'll feel clunky and cryptic.

If you want to see examples of what each of these can do, let me know. I can post some of my own or I'd be more than happy to process some of your photos and send them to you.

Tom
benedict is offline Find More Posts by benedict
Reply With Quote
Reply


Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
Discussion iMax B6AC detecting 2 cells of 3. Buzzer detects all 3. FromHell Batteries and Chargers 9 Dec 01, 2014 12:14 PM
Discussion Watchin' the detectives....... StarHopper44 Trader Talk 32 Oct 10, 2014 04:57 PM
Video Sebart Sukhoi 29s 50e v2 X2 Nir Mazor & Oren Mazor oren 3D Fuel Plane Talk 0 Aug 23, 2011 10:40 AM
Video krill katana 33% Nir-3D oren 3D Fuel Plane Talk 0 Jun 20, 2011 05:00 PM
Berg 5 DSP auto detect not detecting rickyblaze Radios 2 Oct 29, 2004 05:29 PM