[opendtv] Re: "oversampling almost always improves the quality of image acquisition"

From: "TLM" <TLM@xxxxxxxxxx>
To: <opendtv@xxxxxxxxxxxxx>
Date: Wed, 17 Apr 2013 08:10:14 -0700
There have been a bunch of great articles and research papers written about
the various tradeoffs between:

* The market drive to higher sensor resolution (in megapels)
* The drive to smaller (higher yield) sensor chips (in millimeters or
inches)
* The resulting shrinkage of the size of each sensor "pixel" site
* The shutter (capture or integration) time
* The number of photons that can hit that shrinking-sized site in that
sample (time) window at any given light level
* The number of resulting electrons produced in a frame time per site
* The noise floor of the associated electronics
* What the A/D converter has to deal with in order quantify as a result in
that time period

Then you start to think about signal to noise ratio in a whole different
way, especially given the fact that you are dealing with some stochastic
integer number of tens of thousands of photons and electrons per event, not
some analog fluid (liquid) flow of intensity information into the pixel's
charge collection bucket.

So how many bits of precision are needed for each nano A/D and how many bits
are needed at the output of the digital averager?  6-7 bits per nano A/D and
if you were averaging 4 or 9 or 16 sites another 2 - 3 bits out of the A/D?
Or, would you sum the charge from a number of adjacent sites in the analog
domain, sample and hold, and A/D that?  I am not a camera designer but I
*am* a curious electrical engineer.....  I really don't know how this is
done in the real world.  Does anyone on this list have data points or
factoids?

Tom McMahon
Del Rey Consultancy
TLM@xxxxxxxxxx
WWW.DelRey.Com


-----Original Message-----
From: opendtv-bounce@xxxxxxxxxxxxx [mailto:opendtv-bounce@xxxxxxxxxxxxx] On
Behalf Of Craig Birkmaier
Sent: Wednesday, April 17, 2013 6:22 AM
To: opendtv@xxxxxxxxxxxxx
Subject: [opendtv] Re: 4KTV at CES


On Apr 16, 2013, at 6:58 PM, TLM <TLM@xxxxxxxxxx> wrote:

> 
> I know that the lens is symmetrical, but maybe the different types of 
> image sensor chips have differences in their sampling characteristics 
> a la the above?
> 
> Wow, more stuff to keep you up at night...

Some people are making a big deal out of the rapid increase in sensor
densities - 10, 12, 14 MPixels.

What these people are failing to consider is the transformation in sensor
technology from CCDs to CMOS. The two are very different, and decades of
research and development went into the design of CCDs to improve image
quality. CMOS sensors use brute force and averaging to get the job done.
More sensor points allow for the creation of pixels at lower overall
resolution by averaging a bunch of CMOS samples to create one good pixel. 

Resolution can be deceiving -but oversampling almost always improves the
quality of image acquisition.

Regards
Craig 
 
----------------------------------------------------------------------
You can UNSUBSCRIBE from the OpenDTV list in two ways:

- Using the UNSUBSCRIBE command in your user configuration settings at
FreeLists.org 

- By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word
unsubscribe in the subject line.


 
 
----------------------------------------------------------------------
You can UNSUBSCRIBE from the OpenDTV list in two ways:

- Using the UNSUBSCRIBE command in your user configuration settings at 
FreeLists.org 

- By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word 
unsubscribe in the subject line.
[opendtv] Re: "oversampling almost always improves the quality of image acquisition"

Other related posts:
