[opendtv] Re: Image quality
- From: Craig Birkmaier <craig@xxxxxxxxx>
- To: opendtv@xxxxxxxxxxxxx
- Date: Tue, 15 Nov 2005 08:46:28 -0500
At 12:20 AM +0100 11/12/05, Olivier Houot wrote:
>Considering a human eye which is just a 2.5 cm diameter device, using a
>high resolution sensitive region of only 1.5 mm, simple tiny optics
>largely made out of water and jelly ,coupled with motor control and
>processing ressources from just a fraction of the total brain volume,
>manages to do better than high definition, we must clearly be doing
>something wrong...
Not so much wrong, but rather our electronic imaging systems are
rather crude by comparison.
It is very important to consider how the human visual system actually
acquires a high resolution image.
A camera, captures a view of a scene via the lens. Generally speaking
the resolution in that view is relatively consistent across the
image. As we have seen in this thread, lenses may have better
performance in the center than near the edges. One could say that the
human visual system is similar in this respect, however, this is
where the really big difference exists between electronic and human
image acquisition.
A better way to think of this would be a camera designed with two
sensor systems:
One system would acquire a large, low resolution view of the scene
using the entire lens, and gather statistics to drive the second
sensor system. The second sensor would capture a high resolution view
of a small portion of the field of view, using only the best
(central) part of the lens. Based on the statistics from the low
resolution sensor, this camera position would move periodically so
that the central portion of the lens would be able to capture the
high resolution details of interest; these details would be added to
the low resolution image to increase detail in the areas of interest.
At approximately 200 millisecond intervals the camera would would
move to a new spot capture a high resolution view, then move to
another spot, etc. As the camera moved around the view from the low
resolution sensor would change, bringing in more low resolution
information in the area surrounding the high resolution view.
Rather than having a "frame rate, per se' the images would be
constantly refreshed as new information is acquired.
By the way, there are some 3D imaging systems that do something like
this today, using multiple views to build a 3D model of the scene.
The human visual system is quite remarkable. But the thing that
really makes it work is the huge mass of brain cells sitting behind
the sensors that actually process all of the data, directing the
"sensors" to grab additional detail where it is needed.
Acquiring planar images from sensors that are expected to deliver
high resolution across the entire plane is rather crude by comparison.
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.
Other related posts:
