Introduction: DIY Infrared Plant Analysis
Hi! We're part of the Public Lab community, which develops DIY pollution measuring tools. This is a technique for modifying a regular point and shoot digital camera or webcam into a multi-spectral camera -- similar to part of the LANDSAT satellite's camera -- which takes an infrared and normal light photograph at the same time. This video, from an old Kickstarter campaign, explains the overall project:
The basic technique involves removing the built-in infrared-block filter from a digital camera or webcam, and replacing it with a red filter -- currently we're using a Rosco "Fire" #17 theater gel -- or a blue filter (which we're moving away from as it doesn't work in all cameras). After an important custom white-balancing step, images can be uploaded to the open-source, free service at Infragram.org for conversion and analysis. This Instructable sources heavily from the Infragram wiki page on the Public Lab site.
Get some filter material for this conversion in the Public Lab store (where proceeds support the PL non-profit), or order some yourself.
You can learn more about these techniques and become part of the community at Publiclab.org.
Step 1: Introduction
Vineyards, large farms, and NASA all use near-infrared photography for assessing plant health, usually by mounting expensive sensors on airplanes and satellites (see the first image, above). At Public Lab, we've developed a Do-It-Yourself way to take these kinds of photos, enabling us to monitor our environment through quantifiable data.
Our technique uses a modified digital camera to capture near-infrared and blue light in the same image, but in different color channels. (note that the above images show a blue filter, but we're starting to shift to using red filters) We then post-process the image to attempt to infer how much it is photosynthesizing. This allows us to better understand and quantify how much of the available light plants are metabolizing into sugar via photosynthesis. Read more about why this works here.
You can do this yourself (as with all Public Lab tools) but there is also an Infragram DIY Filter Pack available in the Public Lab Store.
Step 2: What Is It Good For?
Multispectral or infrared/visible photography has seen a variety of applications in the decades since it was developed. We have focused on the following uses:
- Taking pictures to examine plant health in backyard gardens, farms, parks, and nearby wetlands.
- Monitoring your household plants.
- Teaching students about plant growth and photosynthesis.
- Creating exciting science fair projects!
- Generate verifiable, open environmental data.
- Checking progress of environmental restoration projects.
- Documenting unhealthy areas of your local ecology (for instance, algal blooms)
Notable uses include this photograph of an unidentified plume of material in the Gowanus Canal (and writeup by TechPresident) and a variety of projects at a small farm in New Hampshire at the annual iFarm event. The Louisiana Universities Marine Consortium has also collaborated with Public Lab contributors to measure wetlands loss following the Deepwater Horizon oil disaster.
Above is an example of what one of our "Infragram" cameras sees (left, using an older blue filter) and the post-processing analysis which shows photosynthetic activity, or plant health (right). This picture was taken from a commercial airplane flight.
Step 3: How Does It Work?
Camera modification: We've worked on several different techniques, from dual camera systems to the current, single-camera technique. This involves removing the infrared-blocking filter from almost any digital camera, and adding a specific blue filter -- and more recently, a red filter; the Roscoe #17 "Fire" filter.
A blue filter filters out the red light (the red filter, conversely, the blue light), and measures infrared light in its place. Read more about the development of this technique here. You can also learn more about how digital camera image sensors detect colors at this great tutorial by Bigshot.
Modifying your camera can be relatively easy or a bit involved, depending on the model of camera you have. There's a growing list of moddable cameras and tips for different models on this wiki page. Many webcams can be easier as they're not as dense or complex as a full point-and-shoot camera.
This video walks through the process of opening and modifying a Canon A490, and while it shows the use of exposed negative film instead of a blue or red filter, the process is the same:
History of the project
While we formerly used a two-camera system, research by Chris Fastie and other Public Lab contributors have led to the use of a single camera which can image in both infrared and visible light simultaneously. The Infrablue filter is just a piece of carefully chosen theater gel which was examined using a DIY spectrometer. You can use this filter to turn most webcams or cheap point-and-shoots into an infrared/visible camera.
Step 4: Background: Satellite Infrared Imaging
The study of Earth's environment from space got its start in 1972 when the first Landsat satellite was launched. The multispectral scanner it carried, like the scanners on all subsequent Landsat satellites, recorded images with both visible and near infrared light. Remote sensing scientists quickly learned that by combining visible and infrared data, they could reveal critical information about the health of vegetation. For example, the normalized difference vegetation index (NDVI) highlights the difference between the red and infrared wavelengths that are reflected from vegetation. Because red light is used by plants for photosynthesis but infrared light is not, NDVI allows scientists to estimate the amount of healthy foliage in every satellite image. Thousands of scientists, including landscape ecologists, global change biologists, and habitat specialists have relied on these valuable satellite-based NDVI images for decades.
There are public sources of infrared photography for the US available through the Department of Agriculture -- NAIP and Vegscape -- but this imagery is not collected when, as often, or at useable scale for individuals who are managing small plots.
Caption: Normal color photo (top) and normalized difference vegetation index (NDVI) image. NDVI image was derived from two color channels in a single photo taken with a camera modified with a special infrared filter. Note that tree trunks, brown grass, and rocks have very low NDVI values because they are not photosynthetic. Healthy plants typically have NDVI values between 0.1 and 0.9. Images by Chris Fastie. Visit the gallery of high-res images by Chris Fastie
Step 5: How to Process Your Images
Post-processing: Once you take a multi-spectral photograph with a modified camera, you must post-process it, compositing the infrared and visible data to generate a new image which (if it works) displays healthy, photosynthetically active areas as bright regions.
Software: We're continuing to work on an easy process to generate composite, infrared + visible images that will reveal new details of plant health and photosynthesis. There are several approaches:
The easiest way is to process your images online at the free, open source Infragram.org, pictured above.
But other options include:
Command-line processing of single images and rendering of movies using a Python script. Source code can be found here
Thanks, and if you have trouble, please consider posting your questions to PublicLab.org!