The standard two-dimensional RGB image has had a great run, but it appears to be reaching the end of its monopoly on our camera technology. In this work, Zhu et al. present a camera that can simultaneously capture all of the rich “plenoptic” information from a 3D scene, along with its multispectral content, within a single snapshot. This includes both the full light field emerging from the sample, as well as nearly 30 spectral measurements per pixel. A machine learning approach helps decode and interpret their high-dimensional dataset into an interesting mix of scene data, which could, for example, both produce a depth map and estimate the material properties of various objects within the image. It will be exciting to see how these extra optical dimensions alter the digital images of the future, as cameras turn more into information-gathering devices and take us beyond the conventional 2D photos that we’re accustomed to.

---

---

 Add Comment

You must log in to add comments.