Image processing of confocal images (flattening)

Hi all,

I need help, I am new in matlab.

I need to done threshold of confocal images but problem is that I have to process them first because they are not well taken; sample was liquid and there is some konvex mistake, edges (somewhwre and half of image) of images are darker and not in same level (height) as a middle of image, so with pure threshold I cant get relevat datas (I have big number of images to process)of surface coverage. I need to get a flat picture of sample on same height as a middle of sample.

I have Matlab, not image processing toolbox. I am interested is it possible, and how this kind of image processing?

here are links for two of my images

http://img813.imageshack.us/i/testt.gif/

<http://www.uploadhouse.com/viewfile.php?id=9395381&PHPSESSID=2c010f135dc2fe8a05537baf4deefed4>

[David Young]

Could you make an example image available please? (See markup help for how you can do this. Also note that you can edit your question rather than post a new question or an answer.)

[Sean de Wolski]

Danijela, you have to post an image as David has requested. There is no way for us to determine what you want otherwise and the infinite monkey is hungover today so your chances with it aren't so good.

[David Young]

Danijela, what you have to do is put your image on the web somewhere. There are sites where you can upload an image and then get a link to it. (I don't know about them myself, but other posters here have used them.) Then you just post the URL in your question (ideally with << >> round it!)

[Walter Roberson]

Sorry, I didn't find a way to get the second image to appear in the original posting. The two images are on different hosting companies and I haven't figured out the trick to directly reference the image on the second company.

Answer 1

[David Young]

I think this is quite difficult. There seem to be two problems with the images: one is the defocusing; the other is the change in illumination (especially visible in the second image).

I am not sure how to deal with the change in focus, but anyway what to do depends on exactly what you need to measure. It may be that the lack of sharpness in part of the image is not a problem, if, for example, you just need to estimate the area of the darker structures. It is possible that deconvolution techniques may be some help, and if the blurring is a major problem then you could investigate that.

The gradient in illumination is easier to deal with, using adaptive thresholding. There's a file exchange contribution which might be applicable, and also a relevant demo with the Image Processing Toolbox.

Try fixing the illumination problems first, and then assess how much the blurring matters. What you do then depends on what exactly you need to get out of the images.

[Danijela]

Thank you very much, I'll try to do it that way and se what will I got.

Answer 2

[Roman Voronov]

i think for the first image it should be pretty easy to segment those large objects using simply im2bw(image,graythresh(image)), despite the focus problem

for the second image there is a pretty similar example here: http://www.mathworks.com/help/toolbox/images/f0-8778.html

however, you do need the image processing toolbox if you are going to do any kind of image processing!

~roman