Scientists at the University of Glasgow in the United States have created a radically new visualization method using artificial intelligence (AI) to transform time into visible three-dimensional space. The new development could help cars, mobile devices, and health monitors develop 360-degree awareness, Optica report.
Photos and videos are usually created by capturing photons – the building blocks of light – with digital sensors. For example, digital cameras are made up of millions of pixels that form images by detecting the intensity and color of light at every point in space.
3D images can be generated either by placing two or more cameras around an object when it is filmed at different angles, or by using streams of photons to scan a scene and reconstruct it in three dimensions. In any case, the image is created only by collecting spatial information about the scene.
Researchers from the UK, Italy, and the Netherlands describe a completely new way of creating animated 3D images: they capture temporal information about photons instead of their spatial coordinates.
Information about the timing of each photon reflected in the scene – what researchers call temporal data – is compiled into a very simple graph. These plots are then converted into 3D using a sophisticated neural network algorithm. The researchers trained the algorithm by showing thousands of ordinary photographs of people moving and carrying objects around the laboratory.
In the end, AI figured out how temporal data correlates with photographs. During the experiment, the team was able to construct moving images at roughly 10 frames per second from temporal data, although the equipment and algorithm used can generate thousands of images per second.