I have in project to build a thermal-IR camera. I know there's already a arduino project to build something like this. What i want is something more in the range of a video camera with thermal-IR, not something that has to scan a area to make a thermal image. I want to it be live. I want to use this to go into the woods and see wildlife.
I was think of using the MLX90620 sensor. I don't know if this is possible. Any input from you guys would be appreciated.
It is not just a matter of wiring up the sensor to get a 16 by 4 pixel image. You have to focus an image onto the sensor. For that you need a lens that works at far IR. That means a germanium or silicon lens and that means big $.
A lot of money for a very small image.
That device is almost certainly a staring array, not mechanically scanned.
It clearly isn't impossible though, because you can buy one.
It is probably also near-IR.
As Mike said, the optics are exotic (most common glass is opaque/reflective to the sorts of wavelengths you're interested in - find some thermal footage on YouTube and try looking into cars) and a 16 x 4 image by itself is very limited, so the scanning mechanism is essential.
Employing either a 320 × 240 or 640 × 480 FLIR thermal imaging core with advanced image-processing algorithms,
So an arduino with a 320 X 240 image will need 76.8K bytes, how is this going to fit into the 2K SDRM?
So building something like that with an arduino is impossible.
I was assuming that you'd be reading and sending a row (or even a pixel) at a time, not as an entire image.
However, there remains the problem of digitising at video rates.
This really ins't Arduino territory, unless you're only using it to drive the scanning mechanism.