Open tonyfast opened 9 years ago
@tonyfast Before trying them out, I wanted to ask if these are relevant to our purpose or not. Are these worth trying? http://www.mathworks.com/matlabcentral/fileexchange/24998-2d-3d-image-segmentation-toolbox http://www.mathworks.com/matlabcentral/fileexchange/34224-fast-continuous-max-flow-algorithm-to-2d-3d-multi-region-image-segmentation
Are any of they really easy to try out?
On Wed, Oct 8, 2014 at 1:37 PM, glahoti6 notifications@github.com wrote:
@tonyfast https://github.com/tonyfast Before trying them out, I wanted to ask if these are relevant to our purpose or not. Are these worth trying?
http://www.mathworks.com/matlabcentral/fileexchange/24998-2d-3d-image-segmentation-toolbox
— Reply to this email directly or view it on GitHub https://github.com/awhite40/MIC-Modeling-Polymer-Composites/issues/9#issuecomment-58396167 .
Tony Fast, Ph.D. tonyfast.com Linkedin http://www.linkedin.com/pub/tony-fast/38/8b0/639 | SlideShare http://slideshare.net/tonyfast1 | GitHub http://github.com/tonyfast
Materials Data Scientist
Woodruff School of Mechanical Engineering
Georgia Institute of Technology
801 Ferst Drive NW
Atlanta, GA 30332-0405 m: 609.509.6492
May or may not be. One of them requires some of the C++ files to be compiled first and then one can try the matlab files. So I wanted to ask- Is there a way to compile C++ files inside Matlab if there is a C++ compiler on the machine?
The other one is giving an error while I was trying to run their demo. It wasn't able to read .nii file.
I will show to you in or after the class if you will be free.
C++ can be compiled by matlab using the http://www.mathworks.com/help/matlab/ref/mex.html function.
Let's talk during class. This may take some work.
Please look into the http://www.bluequartz.net/?page_id=97 also.
@tonyfast http://www.bluequartz.net/?page_id=97 : This seems to be working only for 2D images.
Someone has written down this EM/MPM segmentation algorithm in matlab also.(http://www.mathworks.com/matlabcentral/fileexchange/43410-em-mpm-image-segmentation-algorithm) It is also working only for 2D images.
I have tried both of them.
But we know that theoretically, EM/MPM works for both 2D & 3D images.
How do they work?
On Thu, Oct 9, 2014 at 4:39 PM, glahoti6 notifications@github.com wrote:
But we know that theoretically, EM/MPM works for both 2D & 3D images.
— Reply to this email directly or view it on GitHub https://github.com/awhite40/MIC-Modeling-Polymer-Composites/issues/9#issuecomment-58573812 .
Tony Fast, Ph.D. tonyfast.com Linkedin http://www.linkedin.com/pub/tony-fast/38/8b0/639 | SlideShare http://slideshare.net/tonyfast1 | GitHub http://github.com/tonyfast
Materials Data Scientist
Woodruff School of Mechanical Engineering
Georgia Institute of Technology
801 Ferst Drive NW
Atlanta, GA 30332-0405 m: 609.509.6492
Did you mean how does EM/MPM works? Then, I think these are pretty good references: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=869185
https://engineering.purdue.edu/~ace/thesis/christopher/christopher-thesis.pdf
http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=413651
If you meant something else, then please clarify.
If you haven't, try them out.
Make up some rules
t
to get the code even working, STOP!t
to anything useful for the code, STOP!Don't go in too far.
I have tried them out. They are pretty fast but these available software/matlab code take in only 2D image as an input. So they don't seem to be serving our purpose but may be useful for other people in their projects. But as I said earlier, this method can definitely be used for 3D image but as of now I don't know 'how'. If required, it can definitely be understood and coded.
Trying to answer your questions: Our goal of segmentation endeavor is to isolate fiber and matrix. If that is being successfully done, further statistical analysis can be carried out.
Geet,
The statistics you want should drive the segmentation. Make a list of the statistics you want and decide on what needs to be extracted from the image to proceed.
Like if I want to the length of the fiber as a statistic, I don't need to know all the voxels that make up that fiber. Decide on your methods according to what you need at this point, not what you could get.
On Thu, Oct 9, 2014 at 5:15 PM, glahoti6 notifications@github.com wrote:
I have tried them out. They are pretty fast but these available software/matlab code take in only 2D image as an input. So they doesn't seem to be serving our purpose but may be useful for other people. But as I said earlier, this method can definitely be used for 3D image but as of now I don't know 'how'. If required, it can definitely be understood and coded.
Trying to answer your questions: Our goal of segmentation endeavor is to isolate fiber and matrix. If that is being successfully done, further statistical analysis can be carried out.
— Reply to this email directly or view it on GitHub https://github.com/awhite40/MIC-Modeling-Polymer-Composites/issues/9#issuecomment-58578927 .
Tony Fast, Ph.D. tonyfast.com Linkedin http://www.linkedin.com/pub/tony-fast/38/8b0/639 | SlideShare http://slideshare.net/tonyfast1 | GitHub http://github.com/tonyfast
Materials Data Scientist
Woodruff School of Mechanical Engineering
Georgia Institute of Technology
801 Ferst Drive NW
Atlanta, GA 30332-0405 m: 609.509.6492
Segmentation based on gaussian likelihood maximization:
Two Phases:
Three Phases:
These look great. I imagine they are similar to the results in this post that Uses Gaussian Maximum Likelihood Estimation. This post was generated based on the function
I think y'all are really close to having some working images. Let's plan the road forward soon.
I think we need to use the peaks in the image histogram to better identify the phases. The Peaks of the matrix and fiber in the histogram and heavily convolved and can not be extracted naively.
Things to test: