By using reshape () in Matlab getting this error .

img=imread('flair.jpg');
gaborArray = gaborFilterBank(5,6,39,39);            
featureVector = gaborFeatures(img,gaborArray,1,1);   
img_size=size(img(:));
fv=reshape(featureVector, size(img_size),30);  
out=kmeans(fv,5);               %k-means where k=5 
%Apply histogram of texton map within  each super pixel 
res=hist(out,5);
disp(res)

Error: Error using reshape Size arguments must be integer scalars. Error in texton (line 8) fv=reshape(featureVector, size(img_size),30);

Answer 1

[Stephen23]

This

img_size=size(img(:));

will return a 1x2 row vector equivalent to this:

[numel(img),1]

Then you get the size of the 1x2 vector img_size using

size(img_size)

which will therefore always return the vector [1,2], which is not a scalar, and is thus an invalid input to reshape. Note that the syntax size(X) will always return an 1xN vector, where N>=2.

It is not clear what you are trying to do.

[Javaid Iqbal]

Sir, basically I want to reshape on featurVector the given image size and also multiply with 30 because I'm applying gaborFeature. When I put scaler values in place of size(img_size) then no error.

[Javaid Iqbal]

Please see attachment and I just apply the given code on this but facing this error. Please help me how this will be resolve. Thanks

   % code
img=imread('flair.jpg');
gaborArray = gaborFilterBank(5,6,39,39);            
featureVector = gaborFeatures(img,gaborArray,1,1);   
img_size=size(img(:));
fv=reshape(featureVector, size(img_size),30);  
out=kmeans(fv,5);               %k-means where k=5 
%Apply histogram of texton map within  each super pixel 
res=hist(out,5);
disp(res)

[Stephen23]

Put the new size into one vector:

[size(img_size),30]

it works for me:

fv = reshape(featureVector,[size(img),30]);

As its help shows, reshape does not support both multiple inputs and a non-scalar input at the same time: either all separate inputs need to be scalar (one may be empty), or use one non-scalar input.

[Javaid Iqbal]

Thanks sir for helping me