Adiabatic Camera: Multicolor imaging in the mid-Infrared on Silicon sensors

The mid-Infrared part of the spectrum (wavelengths from 2 to 15 μm) is often referred to as the optical “fingerprint” region since many gaseous molecules and bio-molecules can be identified by their unique and characteristic absorption lines in this region. Moreover, the mid-IR spectral region contains two atmospheric transmission windows (3–5 μm and 8–13 μm), in which the Earth’s atmosphere is relatively transparent. These attributes make color mid-IR imaging of tremendous importance for a wide variety of technological applications ranging from remote gas sensing, explosives detection, cancer diagnosis, autonomous vehicles, surveillance and defense to name a few.


Figure 1 Hyperspectral thermal infrared emission where relative radiance spectra from various targets in the image are shown with arrows. The infrared spectra of the different objects such as the watch glass have clearly distinctive characteristics. The contrast level in the image itself indicates the temperature of the object.

The Invention
We have developed, based on our proprietary nonlinear optical crystal design and system architecture, a new technology capable of providing color imaging in the mid-IR on a cheap color CMOS camera, where the RGB colors in the visible have a one-to-one correspondence to colors in the mid-IR.

Figure 2 color imaged obtained on a standard CMOS camera displaying the two different colors imaged (b) spectra in the mid-IR (blue) curve showing the 2 original mid-IR wavelengths comprised in the image. The red curve show how the 2 mid-IR colors are upconverted, at the same time, to vis-NIR (690 nm and 820 nm) and images on the color CMOS.

The Advantage
To date, the most widely used Infrared cameras in the mid- to long- wavelength infrared region are based on materials such as Indium Antimonide (InSb) and Mercury Cadmium Telluride (MCT). However, these cameras are expensive due to material processing, often require cooling, suffer from noise and poor sensitivity and lack the spatial resolution of their Visible-Near Infrared, Silicon based counterparts. But more importantly they are unsuited for mid-IR color imaging in a single shot. More recently, nonlinear conversion of images has been demonstrated, achieving single-photon sensitivity and high resolution imaging of mid-IR signals with a visible camera.  However, these methods require the implementation of a phase matching compensation technique making these methods unsuited for broad-spectral imaging. Our technology, adiabatic upconversion imaging, frees the conversion approach from these constraints and naturally leads to the single shot capture of a color image in the mid-IR using on a low-cost, high sensitivity, fast, and color CMOS allowing to spatially distinguish spectral components in the mid-IR.

Figure 3- Comparison to State-of-the-art mid-IR imaging methods. Mid Wave IR (MWIR) a.k.a. thermal imaging is the most widely used method for mid-IR sensing.

    MCT and InSb based solutions require cooling, and are severely limited in sensitivity and spatial resolution. Furthermore, the resulting images are integrated over a spectral bandwidth and therefore lack color differentiation. In order to achieve MWIR hyperspectral imaging, spectral differentiation can be obtained using a set of spectral filters, each passing a certain bandwidth, however the images at the different wavelengths are acquired sequentially. On the other hand conventional upconversion imaging addresses some of the thermal imaging limitations by converting the mid-IR radiation to visible-near-IR to be subsequently imaged on Silicon based imager. This method is appealing since these detectors are cheap, fast, efficient, high resolution and color sensitive. This technique however is impeded by the phase mismatch, an inherent aspect of nonlinear frequency conversion processes. The standard nonlinear crystal technology results in efficient upconversion only for a narrow spectral bandwidth. Therefore, a tuning mechanism via angle of incidence (as shown) or temperature needs to be applied to the nonlinear crystal and sequential images. In contrast, Adiabatic Frequency Conversion based upconversion imaging allows efficient ultra-broadband nonlinear conversion capabilities onto a standard Silicon focal plane array.


Sign up for
our events

    Life Science