The International Academy, Research, and Industry Association (IARIA) recently presented the “Best Paper Award” to a team of Rambus researchers for their work on ultra-miniature, computationally efficient diffractive visual-bar-position sensors.
Authored by Mehjabin Monjur, Leonidas Spinoulas, Patrick R. Gill and David G. Stork, the paper, which was presented at SENSORCOMM 2015 in Italy, describes the design and performance of an ultra-miniature lensless computational sensor optimized for estimating the one-dimensional position of visual bars.
“The sensor consists of a special-purpose wavelength-robust optical binary phase diffraction grating affixed to a CMOS photodetector array. This grating does not produce a traditional high-quality human interpretable image on the photodetectors, but instead yields visual information relevant to the bar-position estimation problem,” the article abstract explains.
“Computationally efficient algorithms then process this sensed information to yield an accurate estimate of the position of the bar. The optical grating is very small (120 μm diameter), has large angle of view (140◦), and extremely large depth of field (0.5 mm to infinity). The design of this sensor demonstrates the power of end-to-end optimization (optics and digital processing) for high accuracy and very low computational cost in a new class of ultra- miniature computational sensors.”
It should be noted that Rambus researchers have designed and tested (via extensive simulations) an ultra-miniature lensless sensor for estimating the one-dimensional position of a visual bar throughout a large field of view and regardless of the spectral composition of the bar.
“Our end-to-end design approach led to an optical element (panchromatic binary Fresnel zone plate) that while somewhat complicated in design, is simple to manufacture and mount on a CMOS image sensor,” the authors conclude.
“The signal processing operates on the raw sensor signal (rather than a reconstructed or computed image) and is very computationally efficient. There are a number of directions for future work in end-to- end optimization based on these results, such as extending these methods to other image sensing functions.”