Polarization imaging records the direction that light waves vibrate. This information is invisible to the eye but can help tasks such as material detection and improved scene understanding. The paper presents a compact “lensless” polarization camera that captures four linear polarization measurements from a single photo. The system replaces a conventional lens with a simple optical code and uses computation to recover the polarization images.

The researchers built an optical front end made of two parts: a diffuser and a striped polarization mask. A diffuser is a surface that scrambles incoming light into a distinctive pattern on the sensor. The striped mask filters light according to its polarization direction. Together they imprint both spatial and polarization information onto a plain image sensor in a single snapshot.

Rather than forming a picture directly, the device relies on a reconstruction algorithm that explicitly models how polarization and the diffuser produce the measured pattern. The algorithm uses that physical model to untangle the mixed signals and recover four images corresponding to different linear polarization angles. In short, the optics create a coded measurement and the software decodes it back into polarization images.

This approach matters because existing polarization cameras typically use spatial or temporal tricks to separate polarization signals. Those tricks make cameras bigger, heavier, or require multiple exposures. A lensless, single-shot design can reduce size, weight, and the need for moving parts or multiple frames. The work follows recent lensless imaging ideas, such as DiffuserCam, which also trade a lens for coding optics plus computation.