Dynamic Vision Sensors – A Brief Overview
Dynamic Vison Sensors are Asynchronous imagers. Much like the human eye, they are designed to respond to changes in brightness, with no “Frames” to capture. With DVS, individual pixels independently produce an output only if there is a local change in measured brightness. As such, the image is continuously synthesized form a temporal stream of events, typically in the order of a million events per second (Meps). Therefore, Dynamic Vision Sensors have a much higher temporal sampling of a scene and much lower latency than do conventional frame-based imagers, that typically run at about 30-60 fps.
Another differentiating aspect of Dynamic Vision Sensors is that they have an inherently high dynamic range (about 140 dB). This is because DVS pixels use the logarithm of intensity, instead of absolute intensity, to look for changes in brightness, and determine an event occurrence. Accordingly, this makes them highly suitable for applications where extremely fast object recognition/detection is desired and where a very high dynamic range is needed, as is in automotive/autonomous vehicles ...
Dr. Ziad Shukri, Subject Matter Expect
Ziad Shukri has more than 25 years experience in R&D and manufacturing of semiconductors and thin-films. In 1996 he joined Analogic Corporation to develop and scale up semiconductor manufacturing technology for amorphous selenium direct converters for Flat Panel x-ray imagers, which lead to a successful commercialization of the technology. After that, in 2007, he joined Redlen Technologies where he worked on single crystal CdZnTe detectors development and manufacturing. Thereafter, Ziad joined TeledyneDalsa in 2010, where he worked on custom CCD process development and CCD/CMOS integration into foundry manufacturing in collaboration with Teledyne Digital Imaging. During his work at Teledyne, Ziad also lead several MEMS projects on pressure sensors and accelerometers for major customers in the automotive sector. He Joined TechInsights as a Senior Technology Analyst in January 2020.
Ziad holds a masters in engineering degree in Photovoltaics and a PhD degree on Ternary semiconductor compounds both from McGill University in Montreal. He is the co-author of several technical publications and two patents.