Machine vision, autonomous vehicles, and security systems may soon make use of the polarization camera to solve the challenge of detecting camouflaged objects. The polarization camera offers a new way for humans and machines to see the world. Small polarization cameras could soon be used in embedded vision technology.
How the Polarization Camera Works
Polarization is the direction in which light vibrates, and it offers detailed information about objects. Most cameras only detect the intensity and color of light but can’t see polarization. But cameras that capture polarized light can reveal how light is transmitted and reflected.
A wide range of camera lenses and other add-ons claim to be “polarized,” but most current camera technology is only partially polarized. Current “polarized lenses” only reduce wave directionality to reduce camera glare or to create sharper images, and these partially polarized devices only come in bulky form factors.
Cameras that are designed to detect polarized light can be used to detect material stress, camouflaged objects, or dents and scratches. But, to measure light’s full polarization state, several pictures must be taken along different polarization directions. This requirement is what resulted in the bulky optics. A newer strategy to capture polarized images uses specially patterned camera pixels. However, this strategy could not measure the full polarization state and required a nonstandard imaging sensor.
Embedded Vision To Benefit from New Polarization Camera
Researchers from the Harvard John A. Paulson School of Engineering and Applied Sciences have developed a new polarization camera that can produce more fully polarized images. Scientists and engineers used materials designed at a molecular scale along with a new mathematical understanding of how polarization works to create the new camera technology.
The images allow a user to see light from multiple planes at once in a compact form about the size of a thumb. It uses a metasurface that incorporates subwavelength spaced nanopillars to direct light based on its polarization. The light forms four images that offer a full snapshot of polarization at every pixel. The data collected by this camera can help to reconstruct an object in 3D. Polarization information can be used to estimate depth, texture, shape, and whether an object is natural or synthetic.
The polarization camera may soon be used in a wide range of embedded vision applications. This includes vision systems integrated into autonomous vehicles, onboard planes, or on satellites used to study atmospheric chemistry. Other potential uses include remote sensing, facial recognition, and various machine vision applications.
You may also be interested in reading The Importance of Embedded Vision for Industrial Inspection Purposes.