See the whites of....LEDs

February 15, 2016 | 12:00
At right, details of the LED luminaire processed with the PST algorithm become visible
At right, details of the LED luminaire processed with the PST algorithm become visible
In very well-lit areas, our digital images are often overexposed, washed out with light, the details lost. Such loss of information is critical in image recognition techniques. The PST (Phase Stretch Transform) algorithm , developed at the university of California, considerably improves the processing of overexposed images and allows computers to distinguish details which are lost in standard images. PST is based on a proven Analog to Digital conversion technique for very wide band signals (<10 GHz!) by time-stretching. Now known as “Photonic Time Stretch” it is used in medical imaging.

This research allows obtaining a detailed image of, for example, the internal structure of a LED luminaire (see photo) where the actual image, literally blinded, has no contrast. These results can also be profitably used in astronomy, allowing better discrimination of distant stars, as well as recognition of the texture of near objects, for example in industrial quality control. This algorithm, which can also be used in personal ID recognition by iris or fingerprint scanning, consists of a mathematical operation for identification and extraction of outlines. Another possible area of application lies in autonomous vehicles.

It is exactly such applications which the researchers at UCLA are hoping for in opening their source code to the scientific community.
Loading comments...
related items