Every camera you’ve ever used in your life has a lens that focuses incoming photons on to a light-sensitive surface. But, in the future, cameras might not need lenses at all, and this Bell Labs prototype illustrates how this could be done for cheap.
Technology Review reports the Bell Labs concept, which relies on a method for assembling images called comprehensive sensing (or comprehensive imaging). The basic principle states that any data set with lots of similar measurements will contain a lot of overlapping — and therefore redundant — information. In theory, then, if you take better measurements, you can get substantively similar results more efficiently.
The prototype design consists of two different parts: an aperture assembly and an image sensor. The aperture assembly is an LCD, which allows light to pass through at points of variable size and position. The sensor, in this case, is a single-pixel, three-colour chip.
To take a photo, the camera records the light that passes through a random sampling of points on the aperture assembly, and uses processing to find the correlation between the different data, after which it can reassemble the image.
While this whole process takes a bit longer than we're used to by today's photographic standards -- it needs to do the equivalent of thousands of shutter movements instead of a single shutter -- it's unbelievably efficient when it comes to materials and components. Indeed, the Bell Labs prototype was assembled using off-the-shelf materials.
Imagine a camera without optics and hardly any pixels on the sensor. It could be incredibly cheap. The main problem will be getting the image quality and performance of such a camera up to snuff, which is all too far off. [arxiv.org via Technology Review]