I need a matlab code for automatic extraction of region of interest in an image .i dont require extracting ROI manually by creating mask or by cropping ..the code or algorithm should extract automatically without any manual intervention.
Hi
There is not a uniqe solution for ROI extraction. In fact, it is closely depend on the type and features of your images. Indeed, you need differe ROI extraction method for geographic image and medical images. You should search the papers that worked on the same images as yours.Most of the authors provide the readers with their codes.
Yes, the ROI selection is purely subjective matter. You can get a number of codes in MATLAB for ROI selection.
Best Wishes
Long story Marwa , this is my code I did it for a friend .
Assuming the ROI (interactive selection in matlab GUI ) you do not want then it is heavily subjective and you need to handpick the regions with some level of identification.
CODE:
I=imread('Image of interest');
I2=rgb2gray(I);
I22=edge(I2,'Prewitt');
region = offsetstrel('ball',3,3);
EI = imerode(I2,region);
Fifty_pts = detectHarrisFeatures(I22, 'MinQuality', 0.55);
Fifty_addr=Fifty_pts.Location;
[IndxF,CnF]=kmeans(Fifty_addr,2);
[RowF, ColF]=size(Fifty_addr);
x1=CnF(1,:);
x2=CnF(2,:);
TmpPts= Fifty_addr;
for i=1: RowF
if ((TmpPts(i,1) > x1(:,1)) | TmpPts(i,1) < x1(:,1)) & ((TmpPts(i,2) < x2(:,2)) | TmpPts(i,2) < x2(:,2))
TmpPts(i,:)=0.0;
end
end
TmpP=TmpPts(1:37,:);
TmpGrp=nonzeros(TmpP);
Count2=nnz(TmpGrp);
TTlCnt2=Count2/2;
TmpP=reshape(TmpGrp,[TTlCnt2,2]);
xmin=min(TmpP(:,1));
ymin=min(TmpP(:,2));
xmax=max(TmpP(:,1));
ymax=max(TmpP(:,2));
Width=ymax-ymin;
Len=xmax-xmin;
figure ; imshow(I) ; hold on
plot(TmpP(:,1),TmpP(:,2),'g*');
hold off
roi_of_choice=imcrop(I,[xmin,ymin,Len,Width]);
figure ; imshow(roi_of_choice) ;
Now: it starts with infliction points obtained by - detectHarrisFeatures(I22, 'MinQuality', 0.55);
that gives you coordinates with 55% of the energy .
Next , localize the points say using kmeans( ) clustering.
Then , I would call it handpicking because it is visual location which the ROI of your choice.
See the following links:
http://www.jpathinformatics.org/text.asp?2016/7/1/29/186902
http://www.andrewjanowczyk.com/category/deep-learning/
MATLAB code ROI extraction can be found as a supplement to the paper as well as on the blog.
you can use for a rectangular ROI:
rect = [y,x, haut, larg]; crop = imcrop (img, rect) ; imshow(crop,[])
- Badges
- Science topic
- Similar topics
- Analytical Chemistry
- Extraction