Hey, I have written this code for conversion of plot to rgb image followed by conversion to binary image but I am unable to get the binary image and an error that says-"index exceeds matrix dimension"is being displayed. Can someone pls help me? thks!

% plot r = 1; t=0:0.002:1; a=zeros(500); s=zeros(500); for i=1:500 a(i)= deg2rad(t(i)*90); s(i) = (r*pi*t(i)) / 2; xc(i) = r * sin(a(i)); yc(i) = r * sin(a(i)); x(i) = xc(i) + ( s(i) * sin(a(i))); y(i) = yc(i) - ( s(i)* cos(a(i))); plot(x,y)

end % rasterize it to an 50 row x 100 col image; nRow = 50; nCol = 100; xNode = linspace(min(x),max(x),nCol); yNode = linspace(min(y),max(y),nRow); yInterp = interp1(x,y,xNode,'cubic','extrap'); xi = 1:nCol; yi = round(interp1(yNode,1:nRow,yInterp,'linear','extrap'));

im = accumarray([yi(:), xi(:)],1,[nRow,nCol]); imagesc(im); % define threshold frequency threshold=125; % obtain the image size [M,N,clrs]=size(im); M=10 N=10 clrs=2 %convert the image into type double for matlab operations tempimage=double(im); % create space for the lightened image filteredimage = zeros(M,N,2); % compute the average of all color intensities temp1=(tempimage(:,:,1)+ tempimage(:,:,2))/2; % compare this with the threshold temp2=temp1 > threshold; %convert it to a range from 0-255 filteredimage(:,:,1)=255 * double(temp2);

 filteredimage(:,:,2)=filteredimage(:,:,1);
 %
 newimage=uint8(filteredimage);
 figure
 image(newimage);
 axis off

[Andrew Reibold]

Advice to get more responses on future questions:

1) Concise topic title. "Help convert plot to binary image", or "Why do I get 'index exceeds matrix dimension' ", and then elaborate your question in better detail in the body.

If you ask the entire question in the topic, many people won't read it.

2) Format all code in question so that it is readable

It makes it easier for us to help you! :)

Answer 1

[Geoff Hayes]

Devanshee - please format your above code so that it is readable. Highlight the code portions and press the {}Code button. Use the Preview window to verify that the code is readable.

As well, please include the line number that the error corresponds to. The index exceeds matrix dimension is telling you that you are trying to access data within a matrix using indices that are larger than the dimension(s) of that matrix.

Look at the initialization of im which is a 50x100 matrix, the initialization of tempimage which is img cast as a double, and then

 temp1=(tempimage(:,:,1)+ tempimage(:,:,2))/2;

Knowing the dimensions of tempimage, is the above possible?

[Devanshee Tanna]

Sir, I m not sure of that itself. kindly enlighten me on how I can fix it. I m aware of the definition of the error but am unable to fix it.

[Geoff Hayes]

Devanshee - your code assumes that the tempimage is three-dimensional when it is only two-dimensional. See (and use) Image Analyst's solution to this problem.

Answer 2

[Image Analyst]

Look at the workspace and use the debugger. You'll notice that im and tempimage are grayscale images, not color. You got confused because imagesc() uses some kind of funky color map by default (that's why I hate it and never use it). Use image() or imshow() instead. Then if you want funny colors, call colormap(), like

image(im); % or imshow(im, []);
colormap(jet(256);

Then if you want a color image so that you can average the red and green channels together, you have to call ind2rgb with the desired colormap:

rgbImage = ind2rgb(im, jet(256));

Try that and come back with corrected code if you have any problems.

[Devanshee Tanna]

So do I use imshow()/image() in place of imagesc?? I am sorry i still dint get what corrections to do. If you could explain? Thanks a lot!

[Image Analyst]

Yes, like I showed.

[Devanshee Tanna]

That way I end up getting the plot that I initially converted to an image. I need a binary image

[Devanshee Tanna]

%%Something like this?
% plot
r = 1;
t=0:0.002:1;
a=zeros(500);
s=zeros(500);
for i=1:500 
    a(i)= deg2rad(t(i)*90);
    s(i) = (r*pi*t(i)) / 2;
    xc(i) = r * sin(a(i));
    yc(i) = r * sin(a(i));
    x(i) = xc(i) + ( s(i) * sin(a(i)));
    y(i) = yc(i) - ( s(i)* cos(a(i)));
   plot(x,y)

end
   % rasterize it to an 50 row x 100 col image;
nRow = 50;
nCol = 100;
xNode = linspace(min(x),max(x),nCol);
yNode = linspace(min(y),max(y),nRow);
yInterp = interp1(x,y,xNode,'cubic','extrap');
xi = 1:nCol;
yi = round(interp1(yNode,1:nRow,yInterp,'linear','extrap'));

im = accumarray([yi(:), xi(:)],1,[nRow,nCol]);
image(im);
% define threshold frequency
threshold=125;
% obtain the image size
[M,N,clrs]=size(im);
 M=10
 N=10  
 clrs=2
 %convert the image into type double for matlab operations
 tempimage=double(im);
 % create space for the lightened image
 filteredimage = zeros(M,N,2);
 % compute the average of all color intensities
 temp1=(tempimage(:,:,1)+ tempimage(:,:,2))/2;
 % compare this with the threshold
 temp2=temp1 > threshold;
 %convert it to a range from 0-255
 filteredimage(:,:,1)=255 * double(temp2);
 filteredimage(:,:,2)=filteredimage(:,:,1);
 %
 newimage=uint8(filteredimage);
 figure
 image(newimage);
 axis off

    if true
      % code
    end

[Image Analyst]

I hesitate to give solutions for things that I don't think make sense, and it's often a waste of time when the poster finally realizes it doesn't make sense. What you want to do is to take a plot, create some image with some arbitrary crazy color map and then average the red channel and the green channel then threshold it to get a binary image and replace the red and green channels with this binary, black and white image. What is the point? If you do want a binary image of the plot (for some unknown, unspecified reason), there are a lot more straightforward ways to getting that than to do color averaging.

[Devanshee Tanna]

Sir, I am a beginner trying to create a binary image from a plot that I have created, of an involute gear profile. The ultimate aim is to compare this image with the sample image of a gear and see how involute the latter one is.. Because I am a beginner, I am trying to achieve the output, without realizing how tedious it can get. The intention was not to make it this complex.

[Image Analyst]

But why not just assign pixels to a binary image immediately rather than trying to turn a screenshot, with tick marks and tick labels and other unwanted clutter, into the binary image?

Answer 3

[Devanshee Tanna]

Following is the code that I have written and also, i have pointed where the error is. I shall be grateful if you could help! % plot r = 1; t=0:0.002:1; a=zeros(500); s=zeros(500); for i=1:500 a(i)= deg2rad(t(i)*90); s(i) = (r*pi*t(i)) / 2; xc(i) = r * sin(a(i)); yc(i) = r * sin(a(i)); x(i) = xc(i) + ( s(i) * sin(a(i))); y(i) = yc(i) - ( s(i)* cos(a(i))); plot(x,y)

    end
       % rasterize it to an 50 row x 100 col image;
    nRow = 50;
    nCol = 100;
    xNode = linspace(min(x),max(x),nCol);
    yNode = linspace(min(y),max(y),nRow);
    yInterp = interp1(x,y,xNode,'cubic','extrap');
    xi = 1:nCol;
    yi = round(interp1(yNode,1:nRow,yInterp,'linear','extrap'));

    im = accumarray([yi(:), xi(:)],1,[nRow,nCol]);
    imagesc(im);
    % define threshold frequency
    threshold=125;
    % obtain the image size
    [M,N,clrs]=size(im);
     M=10
     N=10  
     clrs=2
     %convert the image into type double for matlab operations
     tempimage=double(im);
     % create space for the lightened image
     filteredimage = zeros(M,N,2);
     % compute the average of all color intensities
     temp1=(tempimage(:,:,1)+ tempimage(:,:,2))/2;-------> this is the line where the error is showing
     % compare this with the threshold
     temp2=temp1 > threshold;
     %convert it to a range from 0-255
     filteredimage(:,:,1)=255 * double(temp2);

     filteredimage(:,:,2)=filteredimage(:,:,1);
     %
     newimage=uint8(filteredimage);
     figure
     image(newimage);
     axis off