Nuclear Engineering Division

Detection & Diagnostic Systems Multimedia

Compressive Passive Millimeter-Wave Imager

We have applied the state-of-the-art compressive sensing (CS) technology of single-pixel imaging to the acquisition of images using black-body (passive) radiation. This radiation is measured at millimeter wavelengths (MMWs), specifically frequencies of 146-154 GHz, part of the extremely high-frequency microwave electromagnetic band. Detection of MMWs at the fidelity and resolution necessary for imaging has only recently been made possible by recent advances in semiconductor-based detection hardware.

The imager we developed – called the Compressive Passive MMW Imager – permits remote viewing under conditions that interfere with optical and infrared imaging systems. Visible and infrared light has short wavelengths and is scattered by atmospheric particulates, such as dust, smoke, pollution, clouds, fog, and precipitation. The longer wavelengths of MMW radiation pass through particulates with little or no interaction.

The imager replaces the array of pixel-size light detectors in an optical digital camera with a radiometer, in effect a single pixel. The radiometer also lets the imager collect spectral information, effectively combining this with our award-winning Passive Millimeter-Wave Spectrometer for Remote Chemical Detection.

PMMW image of Argonne Building 331

PMMW image of Argonne Building 331 (which formerly housed a research reactor), about 300 m away from the imager. Click on image to view larger image.

Of course, a system that takes a picture one pixel at a time would not be very useful, so we employed the CS techniques developed for optical wavelengths. In optical CS, light reflected from an object is focused on an array of tiny mirrors that each point at or away from a single detector, which records successive measurements of the combined intensity of the mirrors pointing at it. The mirror positions are randomly changed between measurements, sometimes hundreds of thousands of times, to create a series of randomly varying light intensity measurements. Then numerical matrix manipulations are used to reconstruct the image.

The single-pixel technique in effect provides a compression short-cut: instead of measuring the light intensity incident on each of millions of pixels, as in a digital camera, and then discarding unnecessary information to compress the image into a data file of reasonable size, CS limits the amount of information collected in the first place.

But tiny mirrors will not work for MMWs, so we created a mask containing hundreds of apertures, or viewing windows, based on a Hadamard matrix. The mask is larger than the area needed to make a single image measurement, so different areas with different patterns can be “exposed” to mimic the randomly changing mirror positions used in optical systems and allow us to use similar mathematical methods to reconstruct the image.

The Hadamard mask used in the prototype has 1,763 different patterns, each measuring 41 by 43 pixels. Measurements are needed for all 1,763 patterns to reconstruct the image, but estimated solutions for all the patterns can be pre-determined and stored on the computer. In this way, we can “fill in” the missing data and form an image from relatively few measurements.

Compressive PMMW Imager

Compressive PMMW Imager: The key components are (1) the Hadamard mask and its positioning stage, shown just left and above the center of the photograph, (2) the imaging lenses on either side of the mask, (3) MMW radiometer behind the second lens, and (4) the software that controls the data acquisition and incorporates the algorithms for image reconstruction, which created the image shown on the laptop. Click on image to view larger image.

This extrapolation is demonstrated in the video presented here. Note that the image is formed in real time, allowing the user to decide when to stop taking measurements. Only about 100 measurements are needed to recognize the outline of the scissors.

The imager hardware comprises a lens to focus the viewed object or scene onto the Hadamard mask, the mask itself with a positioner to line different patterns up with the exposure window, a second lens to focus the matrix-modulated radiation into the radiometer, and data acquisition and processing hardware. LabVIEW is used to process the raw data, reconstruct the image, and display the reconstruction in real time.


Compressive Passive Millimeter-Wave Imager

Please wait until video loads completely.

Is JavaScript enabled on your browser? In your case the answer is

Closed Captioning Transcript

Live Closed captioning is available for this video.

Problems viewing the video? Contact

Watch this video on YouTube: Compressive Passive Millimeter-Wave Imager

Demonstration of Progressive Scanning of Scissors and Real-Time Image Reconstruction Using Passive Millimeter-Wavelength Radiation
Watch video | Transcript

Last Modified: Thu, November 17, 2016 8:52 AM

For more information:

Nachappa "Sami" Gopalsami
Detection & Diagnostic Systems Dept
Sensors & Instrumentation Section
Fax:  +1 630-252-3250