PGD Home

I have been researching Thermal Imaging Technology recently, and I have decided to resurrect this project.

The previous project was postponed due to the difficulty in using the 16*4 thermopile array to produce a thermal image. I was using pan/tilt servo motors to move the thermopile in steps, and then I was attempting to stitch the images together into one.

The issues I was having were due to, mainly, having no lens on the thermopile to focus it. This meant that the edges of the image didn’t line up properly. Lenses that are usable with thermal imaging equipment (far infrared) are expensive, and using them in this project would have been complex. The lenses I was going use for this purpose were made from Zinc Selenide, or ZnSe.

However, I have recently learned of THIS product, the Sparkfun FLIR Development Kit.. It uses an official FLIR Lepton thermal imaging core to produce an 80×60 image. I intend to use one of these with an arduino to finally produce an affordable thermal imager. The resolution (80×60) will not be very high, but it will be far cheaper than any of the commercial models out there. The FLIR lepton is a good thermal imaging core, and is used in a lot of products, such as the TG165 handheld thermal scanner. It should provide good quality images, if a little low resolution, even in outdoor environments.

In addition, I can use interpolation to increase the resolution without too much degradation in image quality.

The only problem I can see is the RAM requirements. I intend to use an Arduino Nano to prototype this project, eventually moving it to a standalone Atmel Atmega 328 chip. These chips have only 2 KB’s of SDRAM.

The thermal imaging core has a resolution of 80*60, which is 4,800 data points. These are temperature values, ranging from -40 celsius to 65 celsius, which means I can use a byte datatype to store them. Bytes store integer values from 0-255 using just one byte of memory. By adding 40 to the data points coming from the thermal imager and subtracting it during saving and rendering, I can make sure the range is between 0 and 255. That means I can store each frame of data using just 4.8 KB’s, assuming no interpolation. However, I only have 2 KB, so I will need to use an add-on RAM chip, such as the Microchip 2k256, which provides 256 KB’s of additional memory. This should be plenty for my needs. This chip is accessed using an SPI interface.

I intend to use an arduino compatible screen to view the image in real time, and an SD Card module to save images and video in RAW format.