Sunday, April 10, 2011

Samsung Applies for Combined ToF and RGB Imager Patent

Samsung application US20110074989 proposes the integration of ToF IR pixels with RGB or RGB-W ones in a single array. One example of the pixel arrangement assumes quite big structure repeating itself across the array (black area transmits IR light):

The cross-section along the lines A-A is:

ToF portion seems to be based on Kawahito ToF approach with in-phase and out-of-phase integration nodes, controlled by electric current flow, Fig 5A and B respectively:

At the same time RGB-W pixels are protected from IR ToF illumination by a p+ potential barrier:

The sensor seems to be aimed to mobile applications.

Update: A separate Samsung application US20110085063 talks about IR-RGB color filter arrangement and RGB interpolation for that case.


  1. The TOF part is similar but hardly "based on". In fact, and unfortunately, there is substantial prior art to Kawahito's work, although he probably did not know of it at the time.

    In any case, since this work we are pursuing other approaches for RGBZ sensors that may be better for the sensor and the RGBZ ISP.

    Personally, I would not have picked this application publication up as a significant news item, but thanks I guess!

  2. What is the basic principle of Kawahito's approach please ??

  3. @ "What is the basic principle of Kawahito's approach please ??"

    It's based on modulated light source and sensor integrating in-phase and out-of-phase signals in order to determine the phase of reflected light.

    @ "The TOF part is similar but hardly "based on"."

    I was referring to Kawahito approach as a general idea. To that extent Optrima, PMD, Canesta are all using the same Kawahito principle, even though each company has improved it in a different way.

  4. @ISW, better check USPTO and other priority dates before saying which company or university improved what. For example, in Dec 2001 Bamji and Charbon filed what became US 6,515,740. Kawahito has a priority date more than a year later. So, you should NOT call it the Kawahito principle, even tho he came up with a similar idea independently later. You could call it the Bamji/Charbon principle. But, that would also not be quite right because the idea of using electric field to steer collection of carriers has been around longer than that.

    Kind of reminds me of "Tompsett Fill and Spill" in CCDs. Walter Kosonocky actually is the guy with the patent. He was not so happy about it being named after Tompsett despite the fact that Tompsett came up with it independently.

  5. Oops, Nov/Dec 2000 for Bamji and Charbon. Two years plus before Kawahito.

  6. I guess the Bamji and Charbon patent and many others are now owned by Microsoft since they acquired Canesta!

  7. Kawahito patent specifically talks about differential approach and background light rejection. Bamji and Charbon patent is more generic. Strictly speaking you are right, Bamji and Charbon have the priority here. However, Kawahito patent is much more specific and detailed, more "disclosing", so to say. May be this is the reason that for many people this approach is associated with Kawahito name.

  8. I remember from different conferences that it is generally said that Bamji/Charbon copied the PMD principle. Actually, there seem to be many different patent applications for the same idea? A bit confusing to the outsiders ...

  9. I think it is always interesting when different groups come up with the same idea. I can't tell you how many times I thought I came up with some new idea only to find out that someone had the idea already and published it either as a paper, as a patent, or lately, just on the web.
    True for many things, like radio and TV, for example. In this country we still celebrate Columbus Day when it is well known that at least 2 groups discovered North America long before Columbus. Naming isn't always fair (including naming "America" after Vespucci)

  10. The work in EPFL is much ealier, they called the device (in CCD at first) LOCK-in pixel. It's not wise to name the method with the name of a guy. You are really not sure that he was the first to propose this. The world is sometimes so big!

  11. @ "It's not wise to name the method with the name of a guy."

    I do not have a license to officially name anything. I've just used a name of the well known person who did some work in this area. If you happen to use CSEM, EPFL, MESA or whatever other name when referring to the same technique, I have absolutely no problem with it, it's just your personal choice. To me, the name of the person allows to describe his approach in one single word that everybody understands. Please do not look for more in that.


All comments are moderated to avoid spam and personal attacks.