Re: Image analysis

[Macy]

Assuming you converted the .jpg to Test3.bmp, which is about 3.9MB file

If you do the following you should be able to read in a sample image:

myfile=fopen("Test3.bmp","r","ieee-le");
x=fread(myfile);
fclose(myfile);

n=1280*1040;
y=x(end-3*n+1:end);
r=y(1:3:end-2);
rr=reshape(r,1280,1040);
% I use octave 2.1.50. which uses the old gnuplot and this shows the whole image
% gsplot rr(1:10:end,1:10:end);

I got the matrix showing the whole image, and the scale at the bottom. a little 
bit of effort and you can automatically find the scale in terms of indices. 

PS: this was all done by inspecting the .bmp image, there is nothing automated 
about capturing the size, etc, even the fact that the image is saved using 3 
bytes in a row, caught me off guard. I thought the whole image field per color 
was captured, but it appears that the data is R, G, B per pixel, so be careful 
if you apply to other images, look at them first with a hex editor.  The above 
script worked on the sample image you sent.

Regards,
Robert

--- address@hidden wrote:

From: Preeti Gaikwad <address@hidden>
To: Macy <address@hidden>, address@hidden
Subject: Re: Image analysis
Date: Wed, 7 Dec 2011 17:53:57 +0100

Thanks a lot for the interest
I would like to measure this the diameter of the big and small pore(it look
like broken eggs :-)) I  think there is intensity variation in color like
gray and black even if i change in rgb I can see roughly the color
variation. scale is also mention on the image like 10um for 35.5-2.03 and
1um for 60-4-RN.03.
Yes the problem is quite complicated I know I did it manually but in python
they are explaining nice examples as I am not familiar with the same I
stick on octave.....
I donot understand much the given example but got ur point...since you can
put the round and define the size of the same and can get the center and
all for me i hv circles not very nice round and border is gray and center
is black so want to know there size....i m not demanding for precise
measurements...would like to send a simple file to start up...It is
attached with the mail....one test2 file I can measure the pore diameter
and size is displayed now in test 3 I want to do the same exercise for
counting the pore using ocave......could you please let me know how we can
put our scale on the image to get this size.....since I am trying this
first time I dont know how to proceed for the same......thanks for ur help
in advance



On 7 December 2011 17:26, Macy <address@hidden> wrote:

> Wow!
>
> What EXACTLY are you attempting to measure?  In the image labeled,
> 35.5-2.03.tif, there appears to be 'broken' egg shells. Are you measuring
> the holes in those shells, the size of those shells, what?  Plus,
> everything appears to be overlapping!
>
> Not a 'simple' analysis at all, but you knew that already, right?
>
> In the image labeled, 60-4-RN.03.tif, there appears to be 'ruffles', with
> no discernibale edges.  Again, what are you attempting to measure?  Closer
> examination the image shows a finer rendition of the first, coarser images.
> However, there appear to be collapsed larger openings?  For example, near
> the writing at the bottom are some very large areas that looked like they
> once were large, round, but now are collapsed by the morass of finer sized
> 'eggshells'
>
> Just not clear on how you define "success" for counting here.
>
> But, again, very complicated by the overlapping nature of the 3D
> structures.
>
> My approach would be to take a SMALL piece of the image. Decide how, and
> what, it means to measure; and write some sample code?
>
> I have done a bit of similar, but much simpler image processing:
> I wrote octave code that takes a 'round' image, finds its center, maps
> 'average circles extending out from that center, then subtracts that
> average from the original to display disruption. I've attached images of
> those results.
>
> I may be wrong, but I think you will have to 'custom' write your octave
> software, but writing it, you will be much more satisfied with the results,
> plus you can control parameters much better, plus you will UNDERSTAND and
> KNOW exactly what your results mean.
>
> Regards,
> Robert Macy, PE
>
> AJM International Electronics
>  [formerly in Silicon Valley]
> P O Box 74241
> 46007 N 38th Ave, Suite 100
> New River, AZ  85087
> tel: 480 466 0895
> fax: 623 465 2974
>  address@hidden
>
>
>
> = = =
> From:   Preeti Gaikwad <address@hidden>
> To:     Macy <address@hidden>, address@hidden
> Subject:        Re: Image analysis
> Date:   Tue 12/06/11 02:59 PM
>
> THanks a lot to take an interest here i m sending two example
> image.....thanks in advance for ur help...
>
> Attachments             35.5-2.03.tif           60-4-RN.03.tif
>



-- 
preeti gaikwad