From: Dale Eason 
To: James Lerch , ATM List 
Reply-To: Dale Eason 


--0-1783779813-1057357198=:33166
Content-Type: text/plain; charset=us-ascii

A followup.   I forget to add the pixel resolution that I have found.  Assuming
my mirror finding algorithm is good.  My software overlays a circle on the
image and it matches as near as I can tell by eye.  I use and edge finding
algorithm for the bright side.  Then I use a least squares fit for a
circle. It uses about 100 points on the bright edge.

Diameter for all zones except 15 was 312 pixels Zone 15 diameter was 310 pixels.

12.5/312 = .04 inches per pixel

So that can gives us a measurement error of + or -  .02 inches I think.

It can also give us an error of + or - .04 inches based on where we think
the mirror center is.

As Mike said we need more resolution.  My web cam can do up to 640 x 480
but then the software slows down by a factor of about 6.

Dale Eason  wrote: James,
Although I haven't had time to do as thorough an analysis as Mike has here
are some of my results.
I used only the A 15 images and only had time to look at rep 0. I used my
algorithm to find the nulls.  Which is similar to flip and diff. Yours are
on the left and mine are on the right. Like Mike's numbers the center zones
varies the most. My outer zones tend to be small than yours.

If you plot these in excel you will see similar results as Mike's. zone 
RTAFT   D. Eason
1  1.1698    1.563
2  2.0197    2.404
3  2.586     2.747
4  3.0469    3.045
5  3.4462    3.405
6  3.8035    3.846
7  4.1299    4.061
8  4.323     4.367
9  4.715     4.768
10 4.9823    4.968
11 5.2356    5.208
12 5.4772    5.499
13 5.7085    5.649
14 5.9309    5.889
15 6.1452    6.129





---------------------------------
Do you Yahoo!?
SBC Yahoo! DSL - Now only $29.95 per month!


---------------------------------
Do you Yahoo!?
SBC Yahoo! DSL - Now only $29.95 per month!
--0-1783779813-1057357198=:33166
Content-Type: text/html; charset=us-ascii

A followup.   I forget to add the pixel
resolution that I have
found.  Assuming my mirror finding algorithm is
good.  My software overlays a circle on the image and it
matches as near as I can tell by eye.  I use and edge finding
algorithm for the bright side.  Then I use a least squares fit for
a circle.  It uses about 100 points on the bright
edge.
 
Diameter for all zones except 15 was 312 pixels Zone 15
diameter was
310 pixels.
 
12.5/312 = .04 inches per pixel
 
So that can gives us a measurement error of + or
-  .02 inches I
think.
 
It can also give us an error of + or - .04 inches based on where we think
the mirror center is.
 
As Mike said we need more resolution.  My web cam can
do up to 640 x
480 but then the software slows down by a factor of about 6.
Dale Eason
<atmpob{at}yahoo.com> wrote:


James,
Although I haven't had time to do as thorough an analysis as Mike has here
are some of my results.
I used only the A 15 images and only had time to look at rep
0.I used
my algorithm to find the nulls.  Which is similar to flip and
diff.
Yours are on the left and mine are on the right.
Like Mike's numbers the center zones varies the most. My outer zones tend
to be small than yours. 
If you plot these in excel you will see similar results as
Mike's.
zone  RTAFT   D.
Eason1 
1.1698    1.5632 
2.0197  
 2.4043 
2.586     2.7474 
3.0469    3.0455 
3.4462    3.4056 
3.8035    3.8467 
4.1299    4.0618 
4.323     4.3679 
4.715     4.76810
4.9823    4.96811
5.2356    5.20812
5.4772    5.49913
5.7085    5.64914
5.9309    5.88915
6.1452   
6.129


Do you Yahoo!?http://pa.yahoo.com/*http://rd.yahoo.com/evt=1207/*h
SBC" target="new">ttp://promo.yahoo.com/sbc/">SBC Yahoo! DSL - Now only
$29.95 per month!
Do you Yahoo!?
http://pa.yahoo.com/*http://rd.yahoo.com/evt=1207/*http://promo.yahoo.
com/sbc/">SBC Yahoo! DSL - Now only $29.95 per month!
--0-1783779813-1057357198=:33166--

--- BBBS/NT v4.01 Flag-4