Monday, July 06, 2009

Kodak Paper Won Walter Kosonocky Award

Kodak ISSCC 2008 paper won Walter Kosonocky Award For Significant Advancement in Solid-State Image Sensors. The paper is titled:

Low Crosstalk and Low Dark Current CMOS Image Sensor Technology Using a Hole-Based Detector.
Eric Stevens, Hung Doan, Jeffery Kyan, Gang Shi, Jian Wu, Hirofumi Komori, Hiroaki Fujita, Christopher Parks, Cristian Tivarus
Proc. ISSCC, pp.59-61 San Francisco, California USA February 2008.

Congratulations to the whole Kodak team!


  1. Is this the same part that falls short of QE as mentioned earlier in the week?

  2. Yes, this is that same part. However, one has to judge Kodak product performance in light of restrictions the company has. Kodak does not have access to 65nm process, like some of its competitors, and it uses 0.11um for its 1.4um pixel. To cope with the loss of sensitivity, Kodak proposed W-RGB filter and made a nice progress to integrate it in camera processing flow.

    In my view Kodak's fresh approach to overcome its restrictions certainly deserves a creativity award. The outcome of this creativity is another matter.

  3. Is creativity required in switching from vanilla to chocolate? It is difficult to gain much from that.

  4. There were several fine papers over the past 2 years considered for the WKA. The winner is chosen by secret ballot by the twenty or so members of the IISW technical program committee.

    As far as vanilla and chocolate goes, Kodak recognized a considerable improvement in both dark current and cross talk by switching flavors. Full well was also improved. For work probably performed in late 2006 and early 2007, the results were admirable. Personally, I also think the switch was conceptually easy, but who would have guessed that the performance would be so dramatically improved? It is always nice to learn something. We can only hope that the Kodak team remains viable in the next few years.

  5. The filter with clear pixels in it looks like an image processing nightmare to me.

  6. True, there is a lot of problems with it.

  7. Looking at the Kodak's award discussion, I just thought about making an open community poll on the best paper, so everybody can say his/her word. Actually, we can extend the scope and include patent applications and technical white papers (such as Toshiba WDR, Pixpolar, ADVIS and the likes) to compete together with regular IEEE/SPIE papers.

    What do you think about this idea? We can name it "creativity award" or some such.

  8. I'm all for it, but we'd all have to be registered to vote, yes? It seems like you have maybe 10 registered users here.

  9. I don't think the registration is required. The vote is stored somewhere in the browser cookies. So double votes are not allowed, unless anyone tries harder - but I do not believe this happens.

  10. I meant blogger dedicated vote engine works this way, that is.

  11. The WKA award nominations are for any paper presented in any venue, just to be clear. But the award is for significant advancement, not just creativity.

    Managing a meaningful award process takes a lot of planning, a lot of time, and is a lot of work. Just ask Albert Theuwissen.

    Good luck with your creativity award. I look forward to seeing the results.

  12. Actually, I thought to make it in two steps: first to post a call for nominations, let it a week to complete. Then the vote itself for another week, while everybody can see the intermediate results.

    Any suggestion how to make it better?

  13. I guess in hindsight, things are always obvious to some. Like the Egg of Columbus.

    After actually reading the paper, the QE looks to be not much different than the conventional "vanilla" sensor. In fact, peak blue was higher. Only peak red was a little lower, probably because of the well. Like an interline CCD. Crosstalk is quite a bit lower. Perhaps this suggests QE on the 1.4um pixel is limited by the BEOL, as "image sensor" mentions.

    Dark current is very low. Much lower than anything reported to date with plain vanilla. Noise looks to be lower with "chocolate", too. Well well.

    Thanks for the link, image sensor.

  14. > Thanks for the link, image sensor.

    As a matter of fact, Eric Fossum pointed me to this paper award. I just forgot to mention it in the post.

  15. "an image processing nightmare to me ???".

    You are probably not an image scientist.

    Let's not forget who we are talking about here: we are talking about Kodak. If there is anybody in this world that can process images that is Kodak. They probably have an army of image scientists that need some excitement every once in a while.


All comments are moderated to avoid spam and personal attacks.