Help request: Image segmentation loop
in [Matlab]
|
From: Ashish Uthama on 16 Apr 2010 13:45 On Fri, 16 Apr 2010 15:35:41 -0300, Nghia Le <nghiatle(a)gmail.com> wrote: > Hi, what I am trying to do is to identify the tumor, normal, or > background region by using the wavelet transform on each block to obtain > mean and std for each component. After that I will form a ID matrix and > use classify function for computer learning process. I dont see why you cant use BLOCKPROC for this. For example, to obtain the mean of each block: >> I=magic(20); >> blockMeans= blockproc(I,[5 5],@(bs)mean(bs.data(:))) blockMeans = 205.1200 195.9200 196.2800 204.6800 196.6800 204.2800 203.9200 197.1200 203.8800 197.0800 196.7200 204.3200 196.3200 204.7200 205.0800 195.8800 You could replace the third argument with a function handle to perform wavelet transform etc. > Concerning your suggestion, I did try the third loop many times and > trials, but the output for all A(:,:1), A(:,:,2), A(:,:,3)... is the > identical matrix. I am sure there is a simple fix for this, but I do not > know how to do it. I will keep posted. Posting code might help CSSMers point out what might need to be corrected.
From: Nghia Le on 16 Apr 2010 15:29 Yes I also tried this, but the problem is that dtw2 has [cA,cH,cV,cD] components, I am having trouble defining a function handle that will take mean or std for each component by using blockproc. "Ashish Uthama" <first.last(a)mathworks.com> wrote in message <op.va9qda0ua5ziv5(a)uthamaa.dhcp.mathworks.com>... > On Fri, 16 Apr 2010 15:35:41 -0300, Nghia Le <nghiatle(a)gmail.com> wrote: > > > Hi, what I am trying to do is to identify the tumor, normal, or > > background region by using the wavelet transform on each block to obtain > > mean and std for each component. After that I will form a ID matrix and > > use classify function for computer learning process. > > I dont see why you cant use BLOCKPROC for this. > > For example, to obtain the mean of each block: > > >> I=magic(20); > >> blockMeans= blockproc(I,[5 5],@(bs)mean(bs.data(:))) > > blockMeans = > > 205.1200 195.9200 196.2800 204.6800 > 196.6800 204.2800 203.9200 197.1200 > 203.8800 197.0800 196.7200 204.3200 > 196.3200 204.7200 205.0800 195.8800 > > You could replace the third argument with a function handle to perform > wavelet transform etc. > > > > > Concerning your suggestion, I did try the third loop many times and > > trials, but the output for all A(:,:1), A(:,:,2), A(:,:,3)... is the > > identical matrix. I am sure there is a simple fix for this, but I do not > > know how to do it. I will keep posted. > > Posting code might help CSSMers point out what might need to be corrected.
From: Ashish Uthama on 16 Apr 2010 15:22 On Thu, 15 Apr 2010 22:59:04 -0300, Nghia Le <nghiatle(a)gmail.com> wrote: > Hi, I am new to medical imaging analysis and matlab. Here is what I am > trying to solve, hope you can provide me some guidance: > I am trying to segment an image into smaller blocks and analyze data > those blocks individually. For example, I have a 20x20 image > (grayscale) and I want to segment into four blocks and form a > 3-dimensional matrix 5x5x16 that I call 'A' by using for loop and here > is my test code: > > I=magic(20); > for m=1:4; > for n=1:4; > vl=(m-1)*5+1; > vh=(m-1)*5+5; > hl=(n-1)*5+1; > hh=(n-1)*5+5; > > A(:,:,:)=I(vl:vh,hl:hh) > end > end > > Here is my problem, I cannot get A to define in term of m and n. For > example if I want to view block A(m=1 and n=3) it would show up > I(1:5,11:15). I been searching for the correct code two days and yet to > find the solution. Hope you can provide some guidance. Thank you in > advanced. Probably not the most elegant of solutions: I=reshape(1:400,[20 20]); A=zeros(5,5,16); [cGrid rGrid] = meshgrid(1:5:20); for thirdInd = 1:16 A(:,:,thirdInd)=I(rGrid(thirdInd):rGrid(thirdInd)+4,... cGrid(thirdInd):cGrid(thirdInd)+4 ); end
From: Nghia Le on 16 Apr 2010 20:47
Thank you, everything works like a charm. You're the best. "Ashish Uthama" <first.last(a)mathworks.com> wrote in message <op.va9uwbk9a5ziv5(a)uthamaa.dhcp.mathworks.com>... > On Thu, 15 Apr 2010 22:59:04 -0300, Nghia Le <nghiatle(a)gmail.com> wrote: > > > Hi, I am new to medical imaging analysis and matlab. Here is what I am > > trying to solve, hope you can provide me some guidance: > > I am trying to segment an image into smaller blocks and analyze data > > those blocks individually. For example, I have a 20x20 image > > (grayscale) and I want to segment into four blocks and form a > > 3-dimensional matrix 5x5x16 that I call 'A' by using for loop and here > > is my test code: > > > > I=magic(20); > > for m=1:4; > > for n=1:4; > > vl=(m-1)*5+1; > > vh=(m-1)*5+5; > > hl=(n-1)*5+1; > > hh=(n-1)*5+5; > > > > A(:,:,:)=I(vl:vh,hl:hh) > > end > > end > > > > Here is my problem, I cannot get A to define in term of m and n. For > > example if I want to view block A(m=1 and n=3) it would show up > > I(1:5,11:15). I been searching for the correct code two days and yet to > > find the solution. Hope you can provide some guidance. Thank you in > > advanced. > > Probably not the most elegant of solutions: > > > I=reshape(1:400,[20 20]); > A=zeros(5,5,16); > [cGrid rGrid] = meshgrid(1:5:20); > > for thirdInd = 1:16 > A(:,:,thirdInd)=I(rGrid(thirdInd):rGrid(thirdInd)+4,... > cGrid(thirdInd):cGrid(thirdInd)+4 ); > end |