how to run a code regarding regionfill() to obtain the background of an image in matlab 2014 because it is giving error that Undefined function 'regionfill' for input arguments of type 'uint8'.

I am giving the code here.

grayImage = imread('brain_tumor.png');
figure,imshow(grayImage);
%h = impoly;
%position = wait(h);

mask = roipoly(grayImage);
figure, imshow(mask);

BackGround = regionfill(grayImage, mask);
figure, imshow(BackGround);

Sir, I have also attached the image herewith.

Answer 1

[Walter Roberson]

regionfill requires the Image Processing Toolbox, R2015a or newer.

Change your code

BackGround = regionfill(grayImage, mask);

to

BackGround = grayImage;
BackGround(mask) = 0;

with no regionfill() call.

[Debarpita Chaudhuri]

Sir, thanks a lot....your code worked. But I have further question that if I want the Region of interest ie the tumor portion then I am using imcrop() function but it is giving error

Error in imcrop (line 93)
[x,y,a,cm,spatial_rect,h_image,placement_cancelled] = parseInputs(varargin{:});

the code is given below.

% grayImage = imread('brain_tumor.png');
figure,imshow(grayImage);
%h = impoly;
%position = wait(h);

mask = roipoly(grayImage);
figure, imshow(mask);

BackGround = grayImage;
BackGround(mask) = 0;
figure, imshow(BackGround);

ROI_Image = imcrop(grayImage, mask);
figure, imshow(ROI_Image);

Kindly help me Sir.

[Walter Roberson]

It is not possible to crop by an ROI. MATLAB does not support "ragged arrays". You can only crop by a rectangle. The rectangle can be the bounding box of the ROI. regionprops(mask,'BoundingBox') can return a structure that has the information.

[Debarpita Chaudhuri]

Sir, I have tried to change the code according to your suggestion. But I am not getting the ROI or the tumor portion. Infact a blank image is coming. My code is given below.

grayImage = imread('brain_tumor.png');
figure,imshow(grayImage);

mask = roipoly(grayImage);
figure, imshow(mask);

BackGround = grayImage;
BackGround(mask) = 0;
figure, imshow(BackGround);

%h = impoly;
%position = wait(h);

%ROI_Image = imcrop(grayImage, position);
%figure, imshow(ROI_Image);

ROI_Image = regionprops(mask, 'BoundingBox');
figure, imshow(ROI_Image);

Sir, kindly help me.

Regards

Debarpita Chaudhuri.

[Walter Roberson]

blobstats = regionprops(mask, 'BoundingBox');
bb = blobstats(1).BoundingBox;
ROI_Image = grayImage(bb(1):bb(1)+bb(3)-1, bb(2):bb(2)+bb(4)-1);
figure, imshow(ROI_Image);

[Image Analyst]

regionfill() smears in surrounding gray levels towards the interior of the region, so you have varying, interpolated gray levels inside. Walter's original code just sets the interior of the region to zeros, so that's a different operation.

[Debarpita Chaudhuri]

Sir, I am ok with the code which Walter Sir gave me for obtaining the background, however regarding the extraction of the Region of Interest (ROI) which is the tumor portion, is there any way I can obtain it in polygonal shape? Because in the boundingBox the shape is rectangular and it is not exactly the polygonal ROI which I am selecting in roipoly initially. Sir kindly help me. The code is given below.

    grayImage = imread('brain_tumor.png');

figure,imshow(grayImage);

mask = roipoly(grayImage);
figure, imshow(mask);

BackGround = grayImage;
BackGround(mask) = 0;
figure, imshow(BackGround);

%h = impoly;
%position = wait(h);

%ROI_Image = imcrop(grayImage, position);
%figure, imshow(ROI_Image);

blobstats = regionprops(mask, 'BoundingBox');
bb = blobstats(1).BoundingBox;
ROI_Image = grayImage(bb(1):bb(1)+bb(3)-1, bb(2):bb(2)+bb(4)-1);
figure, imshow(ROI_Image);

I have attached the original image and the background and the ROI image using BoundingBox.

[Walter Roberson]

Try regionprops() 'FilledImage' or 'ConvexImage'

[Debarpita Chaudhuri]

Sir, I have got a program online for the operation which I need. However it is working only on .bmp images. Is there any way you can help me to make the program work on .jpg, .png or dicom images. Sir, I would also like to know is there any other simpler code for the same operation. Thank you Sir for your cooperation so far. The program is given below.

if true%--------------------------------------------------------------------------
                    %%%charge et lecture de l'image %%%
%--------------------------------------------------------------------------                
[filename, pathname] = uigetfile('brain_tumor.bmp', 'open image');% only image Bitmap
if isequal(filename, 0) || isequal(pathname, 0)   
  disp('Image input canceled.');  
 X = [];    map = []; 
else
  [X,MAP]=imread(fullfile(pathname, filename));%Lecture d'image.
end;
lsz=length(size(X));
if lsz==3
X=rgb2gray(X);
end
%---------------------------------------------------------------------------------------------------
%%%Region of interest extraction ( Extraction d'une région d'intéret à main libre )by free hand %%%
%---------------------------------------------------------------------------------------------------
im0=X;
X=double(X);
h_im = imshow(im0),title('Choose ROI from the figure');
%helpdlg('Choose ROI from the figure','Point Selection');
h = imfreehand;
position = wait(h);
e = impoly(gca,position);
BW = createMask(e,h_im);
ROI=X.*BW;
figure,imshow(ROI,[]);
im_DIF=X-ROI;
X=X-ROI;
figure,imshow(im_DIF,[]);
ROI=uint8(ROI);im_DIF=uint8(im_DIF);
imwrite(ROI,'Region of interest.bmp','bmp');
imwrite(im_DIF,'Image difference.bmp','bmp');
  % code
end

Sir, I have attached the images herewith. Kindly help me Sir.

[Walter Roberson]

Jpeg images should seldom be used for this kind of work. Jpeg blurs sharp edges making edge detection more difficult.

[Walter Roberson]

That code prompts for the image to open. You can select any of a number of image file formats including jpeg and png. Jpeg just won't do a good job.

Dicom would require a few more lines to read in.