| |
 |
|
|
Science Forum Index » Image Processing Forum » Advise on object detection...
Page 2 of 2 Goto page Previous 1, 2
|
| Author |
Message |
| serg271... |
 Posted: Sat May 24, 2008 8:37 pm |
|
|
|
Guest
|
On May 25, 7:22 am, ImageAnalyst <imageanal... at (no spam) mailinator.com> wrote:
Quote: On May 24, 12:54 pm, serg271 <serg... at (no spam) gmail.com> wrote:
I suspect internal camera format is YUV, before it converted to RGB
later by driver - it seems it's either yuv or jpeg. So working with
asymmetrical YUV could be preferable.- Hide quoted text -
----------------------------------
Why do you say that? Nearly all CCD's I see are natively RGB, and
most (all?) CMOS sensors are CMY or CMYG. Those are the filters that
are actually over the photosites. (Exceptions being the Foveon sensor
(sort of), and the Fuji SuperCCD SR and the new Kodak sensor with
additional "clear" pixels.)
Regards,
ImageAnalyst
One thing I'm sure of - input for driver is not native RGB, because
conversion to RGB done in software and take some time. |
|
|
| Back to top |
|
| serg271... |
| Posted: Sat May 24, 2008 8:47 pm |
|
|
|
Guest
|
On May 25, 9:37 am, serg271 <serg... at (no spam) gmail.com> wrote:
Quote: On May 25, 7:22 am, ImageAnalyst <imageanal... at (no spam) mailinator.com> wrote:
On May 24, 12:54 pm, serg271 <serg... at (no spam) gmail.com> wrote:
I suspect internal camera format is YUV, before it converted to RGB
later by driver - it seems it's either yuv or jpeg. So working with
asymmetrical YUV could be preferable.- Hide quoted text -
----------------------------------
Why do you say that? Nearly all CCD's I see are natively RGB, and
most (all?) CMOS sensors are CMY or CMYG. Those are the filters that
are actually over the photosites. (Exceptions being the Foveon sensor
(sort of), and the Fuji SuperCCD SR and the new Kodak sensor with
additional "clear" pixels.)
Regards,
ImageAnalyst
One thing I'm sure of - input for driver is not native RGB, because
conversion to RGB done in software and take some time.
Confirmation : I have printed values - it's YUV420 - U and V values
are blocks 2x2.
May be camera hardware process sensor format to YUV, or may be driver
doing RGB->YUV->RGB.
I wouldn't be surprised by any suchlike stupidity - it's a Symbian OS
we are talking about. |
|
|
| Back to top |
|
| ImageAnalyst... |
| Posted: Sun May 25, 2008 4:50 am |
|
|
|
Guest
|
On May 25, 2:47 am, serg271 <serg... at (no spam) gmail.com> wrote:
Quote: On May 25, 9:37 am, serg271 <serg... at (no spam) gmail.com> wrote:
On May 25, 7:22 am, ImageAnalyst <imageanal... at (no spam) mailinator.com> wrote:
On May 24, 12:54 pm, serg271 <serg... at (no spam) gmail.com> wrote:
I suspect internal camera format is YUV, before it converted to RGB
later by driver - it seems it's either yuv or jpeg. So working with
asymmetrical YUV could be preferable.- Hide quoted text -
----------------------------------
Why do you say that? Nearly all CCD's I see are natively RGB, and
most (all?) CMOS sensors are CMY or CMYG. Those are the filters that
are actually over the photosites. (Exceptions being the Foveon sensor
(sort of), and the Fuji SuperCCD SR and the new Kodak sensor with
additional "clear" pixels.)
Regards,
ImageAnalyst
One thing I'm sure of - input for driver is not native RGB, because
conversion to RGB done in software and take some time.
Confirmation : I have printed values - it's YUV420 - U and V values
are blocks 2x2.
May be camera hardware process sensor format to YUV, or may be driver
doing RGB->YUV->RGB.
I wouldn't be surprised by any suchlike stupidity - it's a Symbian OS
we are talking about.- Hide quoted text -
- Show quoted text -
-----------------------------------------------
I believe the native signal is RGB because those are the color filters
over a CCD. YUV is a derived signal from that, according to CCIR 656
and CCIR 601 which define standards for component video. I have one
Sony camera where you specify which format you want the video in and
it gives several formats (RGB, YUV, etc.) and several image matrix
sizes to choose from (which implement binning I imagine).
More on the YUV/RGB issue at FourCC.org:
http://www.fourcc.org/fccyvrgb.php
So if you want RGB I believe it comes directly from the camera without
that intermediate conversion like you said. Plus I'm pretty darn
certain that the demosaicing is done in the native color space
(learned this directly from Dalsa's lead vision chip engineer during a
short course). So even after demosaicing it's still RGB (or CMY in
the case of CMOS sensors) so, to me, it makes sense that the format
remains in RGB if RGB is requested and only undergoes the CCIR
specification conversion to YUV if YUV is requested.
Regards,
ImageAnalyst |
|
|
| Back to top |
|
| |
Page 2 of 2 Goto page Previous 1, 2
All times are GMT - 5 Hours
The time now is Sat Oct 11, 2008 1:12 am
|
|