Monday, February 15, 2016

EETimes: Imaging Revolution - Forget Frames

EETimes: "Two French scientists, Ryad Benosman and Christoph Posch... founded Chronocam in 2014. They’ve developed an image sensor designed to capture images not based on artificially created frames, but driven by events within view.

“Each pixel individually controls its sampling – with no clock involved – by reacting to light, or changes in the amount of incident light it receives,” explained Posch, Chronocam’s CTO.

“Frame-based video acquisition is fundamentally flawed,” Chronocam CTO Posch decreed.

Chronocam has applied and tested its sensor’s principles in restoring people’s vision at Pixium Vision, a retina prosthetic company founded by Chronocam’s co-founders.

Chronocam has raised 1.5 million euro so far. Among investors are CEA Investment and Robert Bosch VC.

Chronocam’s CCMA ATIS 1.1 sensor, whose supply voltage is 3.3V (analog), 1.8V(digital), comes in a 9.9 x 8.2mm2 chip size, featuring 2/3-inch optical format. Its array size is 304 x 240 QVGA, with a pixel size of 30μm × 30μm. The power consumption is less than 10mW.

Asked about the company’s next steps, Verre, the CEO, said, “We aim at decreasing pixel pitch and increasing resolution. Our next steps are the VGA tape out and the migration to a CIS process.”

Chronocam is open to licensing its technology to others.
"

How Chrononcam's image sensor sees motions of hand (left)

19 comments:

  1. QVGA, 30 um pixel, 30mW, well, we can all quit working now.

    ReplyDelete
  2. How is this different from Tobi Delbruck's DVS? I think this revolution has been underway for a decade or more...

    ReplyDelete
    Replies
    1. It is not different at all, it is completely based on the work of Tobi's group. And that is not mentioned in a single word in the article, which I find pretty outrageous... They added a second part to the pixel to measure the intensity in a complex way, that's why the pixel is so big.

      In fact there is an ETH spinoff that has been selling such sensors for over 7 years:
      www.inilabs.com
      Recently two more startups have been founded trying to commerzialize the technology:
      www.insightness.com
      www.inivation.com
      So if that Yole guys thinks that Chronocam is unique then he did a bad job...

      Delete
    2. If its completely based on someone else's work, why doesn't this someone sue Chronocam?

      Delete
    3. Eh, for what? Do they have any money that would be worth lawyers' time? Can one demonstrate any damages to their business? You don't get sued for talking, usually.

      Unfortunately even Tobi's work seems to be stuck in the academic phase, not quoite able to find the killer application. Some interesting ideas do pop up, but they themselves tend to be solutions looking for a problem.

      Delete
  3. I found this on youtube:
    https://www.youtube.com/watch?v=XaV_ZP1CQ1c
    Is there any link?

    ReplyDelete
  4. The founder of Choronocam in their IEEE Spectrum gave credit to tobi Delbruck and many others in the field for their work http://spectrum.ieee.org/biomedical/devices/how-neuromorphic-image-sensors-steal-tricks-from-the-human-eye.
    This article is only about Chronocam a company that is targeting the development of neuromorphic cameras and processing, with a time based camera and new algorithms.
    Many people out there designed or tried to design event based cameras, and before Delbruck there was Kramer and before Carver Mead and Misha Mahowald.
    Not surprisingly people commenting on this post are founders of a competing company targeting the same goal, a bit of bashing is always soooo tempting. They’d better focus on developing their technology and compete on the terrain.
    There so much space to operate in this new field, competing companies should better be happy that this technology is talked about.
    Now talking about the technology Chronocam is unique, its cameras with its asynchronous time coded grey scales is one of a kind, competing technology lose the benefits of event driven acquisition by generating conventional frames.

    ReplyDelete
    Replies
    1. The special success of Chronocam is getting media attention.
      It will take a lot more than just a sensor chip to get this approach adopted. There needs to be a killer app (what is it?) with this technology being COMPELLING for the application. Furthermore, I suspect there is a lot more "whole product" that needs to be developed to make it competitive with the large infrastructure already in place to support vision using frame readout sensors. In the technology adoption curve, I don't think the technology has even made it to the enthusiast/innovator class of customer yet. (see technology adoption rate literature)

      Delete
    2. What's the killer app for QIS?

      Delete
    3. I think you should ask about the killer app for photon-counting image sensors, in general, and then about QIS. If I knew the answer, I would have already started a company, ha, maybe QIScam. But, I don't think anyone doubts that photon-counting image sensor technology will have value in the future but aside from low light photography, scientific, and defense applications, the role in consumer electronics and a killer app is unclear.

      Delete
  5. for displaying on a monitor however, the sensor signal most likely, has somehow to be “divided” into frames.

    as a highspeed cinematographer it would be interesting to know, what maximum temporal frequency could be achieved with this technology. in theory, if sampling is quick enough it could be an interesting technology for highspeed cinematography.

    would be interesting to hear your opinion on this possible application for this technology.

    ReplyDelete
  6. My opinion is that using conventional photodiode in this detecting mode can not give high sensitity and good quality image. The best photon sensing is still pinned photodiode + TX readout. We have developped 3T-similar logarithmic pixel with in-pixel differentiation (it has been made just for fun !) as shown a link posed by Anonymo. I agree with Eric, killer applications are hard to find !

    ReplyDelete
    Replies
    1. I think the way this usually works is that killer applications find you, if you happen to be in their path ready to meet them. Looking for them tends to be less rewarding.

      Delete
    2. Yes, you can find the work from Rémi Barbier of IN2P3 in France. He has done this by using his EBCMOS ...

      -yang ni

      Delete
  7. There is also concept of "infinite resolution sensor" which captures the parameters of every photon which hits sensor surface and provide the infinite stream of data in format photon descriptors (wavelength, energy, polarization and of course coordinates X,Y with double precision float number). Has anyone heard of that?

    ReplyDelete
    Replies
    1. Yes, you can find the work from Rémi Barbier of IN2P3 in France. He has done this by using his EBCMOS ...

      -yang ni

      Delete
  8. I uploaded a new video:
    https://www.youtube.com/watch?v=9AD8WfFKWhY
    Does any one can tell what we can do with please ? This has been recorded years ago and I think about it, triggered by this discussion ...

    Thanks !

    ReplyDelete
    Replies
    1. this is way cool! Is there more info on this in public domain? Your NSC1001 flyer gives no indication of this capability

      Delete
    2. This feature is not announced officially but it's implemented in the demo camera. You can buy a such camera to see what you can do with. We have demonstrated a 3D camera during Vision Stuttgart by using such feature which can confine the stereo matching only on the area where there are changes.

      Delete

All comments are moderated to avoid spam.