Tuesday, June 15, 2010

Forza Reviews HDR Approaches

Laser Focus World published Forza Silicon overview of DR enhancement techniques in image sensors. There is a lot of different approaches there and it's hard to cover all of them in a short article, but Forza did a good job selecting a nice subset and describing it in simple words:


Thanks to B.M. for sending me the link!

11 comments:

  1. Lofic technique is the same as multiple exposure with only 2 captures. The SNR would be ~ the same (I bet a bit better for 2 exposures). I really do not understand the utility of LOFIC!!!

    ReplyDelete
  2. I don't think you understand LOFIC. The operation is the same as what I called "cascaded integration" as described in US patent No. 6,888,122. (filed and reduced to practice in 2001 but Micron did not allow me to publish this...) The description there might be helpful.

    http://www.ericfossum.com/Patents/pat6888122.pdf

    Anyway, Forza correctly describes the pros can cons.

    ReplyDelete
  3. I should also mention that probably the Japanese guys independently came up with the same concept. By that time PPD devices were being made more regularly and they were the first to use cascaded integration with a PPD. I was glad to see the idea put to use in a product. I would be more glad if someday they acknowledge the prior Micron work.

    ReplyDelete
  4. the lateral oveflow is different to multple slope where the over-flow charge is lost. In LOIC, this oveflow charge is collected into a second capacitor for form a low gain signal.

    So you can see that there does be a time difference between the high gain channel and the low gain channel, because the low gain channel produces signal only when the high gain channel has overflow. But most of applications, this time difference is largely tolerated.

    Again in this approach, the key issue is to control the overflow threshold and put a large integration capacitor in the pixel.

    TI Japan uses the same principle, I think.

    Y.N.

    ReplyDelete
  5. What about a layer of some photochromic material on top of the sensor?

    ReplyDelete
  6. Photochromic material has a very low reaction time and the attenuation is not very high indeed.
    Y.N.

    ReplyDelete
  7. One of Mizobuchi's LOFIC papers is here:
    http://www.imagesensors.org/Past%20Workshops/2009%20Workshop/2009%20Papers/057_paper_mizobuchi_ti_lofic.pdf

    ReplyDelete
  8. What imagers out there use the LOFIC technique?

    ReplyDelete
  9. I think that the TI 800x600 WDR CMOS image sensor uses this technique. So 2 external ADC used to digitalize 4 images per frame (HG FPN, HG image, LG PFN and LG image), a DSP processes these images to make out a raw image.
    Y.N.

    ReplyDelete

All comments are moderated to avoid spam.