Plant roots detection in neutron imaging

Hi all, I have neutron imaging of soil with plant roots and I need to detect the roots. Problem is that the background is varying according to water content in the soil. I tried thresholding and various blood vessel detection codes but none works well. Does someone have a suggestion? I attached an image for example.

Thanks in advance

Answer 1

[Image Analyst]

Try this http://www.visionbib.com/bibliography/applicat841.html where you'll find papers such as

Zeng, G.[Guang], Birchfield, S.T.[Stanley T.], Wells, C.E.[Christina E.],
Detecting and Measuring Fine Roots in Minirhizotron Images Using Matched Filtering and Local Entropy Thresholding,
MVA(17), No. 4, September 2006, pp. 265-278. 

Kumar, P.[Pankaj], Cai, J.H.[Jin-Hai], Miklavcic, S.[Stan],
Root crown detection using statistics of Zernike moments,
ICARCV12(1130-1135). 

Automated Detection of Root Crowns Using Gaussian Mixture Model and Bayes Classification,
DICTA12(1-7). 

Perciano, T.[Talita], Hirata, R.[Roberto], de Castro Jorge, L.A.[Lúcio André],
Ridge Linking Using an Adaptive Oriented Mask Applied to Plant Root Images with Thin Structures,
CIARP10(378-385).

and several others

[Sharon Dabach]

Thanks, I will contact the authors. However, I was aiming more in the direction of an existing code.

[Image Analyst]

It's so specialized that it's doubtful there is MATLAB code floating around free for the picking that does root finding. I would think root imaging would be fairly similar to blood vessel imaging or other ridge finding methods in general but you said you tried several of them and they didn't work. You didn't provide an image so all I could think of was to find people who have done what you're trying to do.

[Sharon Dabach]

I think the problem with my images is that the background is not uniform. Since I'm not a mathematician, I don't really understand what they talk about in those papers you suggested :( I attached an image to my original question.

[Image Analyst]

Does that image change with the amount of water in the soil? if not, just monitor the mean gray level day by day and that will correlate nicely with root growth. You don't really need to trace out individual roots, I don't think. Just sum up the total amount of roots, which is given by the change in the mean gray level.

[Sharon Dabach]

Yes the image change with water amount in the soil. The more water the less contrast between soil and roots. And I need to know where the roots are So I need a binary matrix for root position.

[Image Analyst]

And were the authors responsive?

[Sharon Dabach]

Yes,, but they don't have matlab codes (only C++).

[Image Analyst]

I guess you'll have to translate it, if they will share it with you. Of course you'd have to offer to supply them with the MATLAB code in return, so try to make it as general purpose and well commented as possible.

[remaoun salim]

Hello guys, I have the same problem, can you help me !!!????

Answer 2

[remaoun salim]

Hello guys, I have the same problem, can you help me !!!????