【電気工学】デジタルカメラ写真を使って隠れた物体を可視化する

Ordinary digital cameras can be used to view out-of-sight objects, reports a paper published in Nature this week. This feat has previously only been achieved with specialist and expensive optical systems, and this cheaper, simpler technique represents a major advance in viewing hidden objects.

Non-line-of-sight imaging techniques analyse the light reflected off a surface - which plays a role similar to that of a mirror in a conventional periscope - to computationally reconstruct the image of an object hidden from view. However, previous techniques have required pulsed lasers and very fast light detectors, at high cost.

Vivek Goyal and colleagues demonstrate a computational periscopy approach using an ordinary digital camera controlled by a computer. Out of sight of the camera, an image displayed on a screen is partially obscured by an opaque object. This occluding object casts a partial shadow (penumbra) on a reflective wall. The digital camera captures snapshots of the distribution of light from the reflective wall, which is made up of light emanating from the hidden screen as well as the penumbra of the occluding object. Computer algorithms then reconstruct the snapshots to create a two-dimensional colour image of the hidden scene. The authors present reconstructed images including cartoon faces, letters and striped patterns, in which larger features (such as white and red patches) are clearly displayed whereas smaller features (such as eyes) are visible but less accurately reconstructed.

These findings demonstrate the feasibility of non-line-of-sight imaging using digital cameras and could have applications in the monitoring of hazardous environments, navigation and detecting hidden adversaries.