John modified a point-and-shoot camera to capture near infrared images. This is half the puzzle in building NDVI images for analyzing crop health.
Information on the conversion can be found here: http://www.lifepixel.com/
You need to be a member of diydrones to add comments!
Comments
David,
We are using Rosco #2007 VS
http://www.rosco.com/filters/cinegel.cfm?CategoryID=33#C2007
The results from unexposed film is a little different from what is being offered on the kickstarter which is along the lines of the SuperBlue filter sold by LifePixel. Unexposed film passes mostly near-infrared light with very little visible light getting through. The SuperBlue and kickstarter filters pass blue and near-infrared light. Having that visible band/channel lets you create NDVI-like images from a single image since the reflectance of both blue and red light by plants is very low. To make NDVI image using unexposed film you need two cameras - one for near-infrared and one for visible (usually red).
Yes, we got CHDK working on it. This tool is very helpful.
http://zenoshrdlu.com/stick/stick.html
David,
Sorry for the delay, I spaced it out. Canon PowerShot A2300 HD :-)
Let my clarify then, please make a blog post about your project as you move forward, your project looks really interesting. I think a lot of people will want to know more about it. Thanks for sharing it!
:-)
I thought that is what I was doing :) but yes of course.
The hope for such a tool like AgPixel is that people play with the data and learn new insights into the remote sensing.. instead of worrying about some sort of consumption based licensing model where you have to pay each time you want to play with new captured data. As much fun as 'cloud' services are.. every business seems to want to get people hooked up on them even when it doesn't make sense.. like here where each RAW file is huge and you may capture a hundred or so when are you out flying... that's a lot of data to upload to the cloud.. and a lot of end users might not have the best internet connection as well... like rural precision farmers.
Jumping back to Gary's comment.. CCD these days have a pretty good sensitivity range for NIR but I would stay closer to red edge .. the farther you go away from the visible light the more distortions you add especially since the optics and coatings of the camera are designed around focusing visible light.. so you likely will risk getting some pretty blurry NIR which makes the usefulness of the NDVI data not as good. If you modified a DSLR there is a $6K lens that has the proper coatings to focus even the NIR. :)
I have been spending quite a bit of my time and resources in remote sensing especially with vegetation indices, working on a commercial toolset (with the help of amazing experts at Kansas State University) that will be out later this summer called AgPixel...here are a couple screenshots from it.. the first is an analysis of the composite image while the second is a NDVI. I have 11 algorithms for vegetation analysis... there are lots of really awesome things you can remote sense.. and you don't need a NIR camera for some of it.
For the NDVI I am using a color pallet where green is highest VI and red is your more stressed vegetation. Personally I like this color pallet as the green = healthy and red = stressed.. the more traditional color pallet where red is the healthy plants confuses me :) The above VI image is stretched using the auto limits in the tool.. the above is a +/- 2*Std Dev stretch which provides a better range to identify things in the image. Images like this is just eye candy for helping identify problematic areas.
The AgPixel toolset is very much a work in progress. Right now I am working on atmospheric corrections using reflectance panel data.. and also supervised and unsupervised classifications.
Reflectance panels are tricky as they need to be a lambertian surface.. where no matter the angle of the sun you get the same reflectance data... a photography gray card is not such a surface. Also I personally wouldn't have a white reflectance panel... I think it is better correct up to white than to be at the extremes already.
It has been a wild ride and amazing to see soo much interest in remote sensing as of late. I'll have to write a blog article in a few weeks when I have more to show/share.
Ned, that looks interesting. I'll pass that on to the nerds on the other side of the office ;-)
Joshua - You might want to look into using the software that uses ImageJ or Fiji to automate the creation of NDVI and false color images. Here is a somewhat outdated note about that software: http://publiclaboratory.org/notes/nedhorning/11-1-2012/update-photo...
I submitted an update to the GitHub repository (https://github.com/nedhorning/PhotoMonitoringPlugin) yesterday to add some additional functionallity including two new plugins the process the SuperBlue filer that is being used in the Kickstarter.
-
1
-
2
of 2 Next