how to fix in imresize error ??

i try to make the function for resizing image, this the code:

>> a=imread('1.png');

>> b=20000;

>> l=sum(sum(a));

>> [m,n]=size(a);

>> m1=fix(m*sqrt(b/l));

>> n1=fix(m*sqrt(b/l));

>> c=imresize(a,[m1 n1]);

but, i get the error :

Error using imresize>scaleOrSize (line 382) Invalid scale or size input argument.

Error in imresize>parsePreMethodArgs (line 354) [scale, output_size] = scaleOrSize(next, next_arg);

Error in imresize>parseInputs (line 248) [params.A, params.map, params.scale, params.output_size] = ...

Error in imresize (line 141) params = parseInputs(varargin{:});

i don't understand what cause of the error and i try to fix the error but i can't fix it. Please Help, Thank you

Answer 1

[Image Analyst]

It's probably color. To fix, try this:

 [m, n, numberOfColorChannels]=size(a)

And use descriptive names like rows, columns, grayImage, etc. not an alphabet soup of names like a,b,c,m,n,i,j, etc. That makes it hard to read code, especially when it's not commented.

[ElizabethR]

thank for your suggestion. I already try but, i still getting error :(

>> a=imread('1.jpg'); ( binary Image )

>> [m, n, numberOfColorChannels]=size(a);

>> WideObject=sum(sum(a));

>> ScaleFactor=20000;

>> m1=fix(m*sqrt(ScaleFactor/WideObject));

>> n1=fix(n*sqrt(ScaleFactor/WideObject));

>> c=imresize(a,[m1 n1]);

[Jialiang Shen]

I think it is due to the scale factor

[Image Analyst]

No, not due to scale factor because if it were truly a binary image, it works. See:

grayImage=imread('cameraman.tif'); % Gray scale image
% Create a binary image
a = grayImage > 128;  % Now the badly-named "a" is a binary image
[m, n, numberOfColorChannels]=size(a)
WideObject=sum(sum(a));
ScaleFactor=20000;
m1=fix(m*sqrt(ScaleFactor/WideObject));
n1=fix(n*sqrt(ScaleFactor/WideObject));
c=imresize(a,[m1, n1]);

The above code works just fine.

However if it's an RGB image, it gives the exact error message she got. See:

a=imread('peppers.png'); % Read in a color image.
[m, n, numberOfColorChannels]=size(a)
WideObject=sum(sum(a));
ScaleFactor=20000;
m1=fix(m*sqrt(ScaleFactor/WideObject));
n1=fix(n*sqrt(ScaleFactor/WideObject));
c=imresize(a,[m1, n1]);

m =
   384
n =
   512
numberOfColorChannels =
     3
Error using imresize>scaleOrSize (line 405)
Invalid scale or size input argument.
Error in imresize>parsePreMethodArgs (line 377)
[scale, output_size] = scaleOrSize(next, next_arg);
Error in imresize>parseInputs (line 260)
[params.A, params.map, params.scale, params.output_size] = ...
Error in imresize (line 146)
params = parseInputs(args{:});
Error in test (line 7)
c=imresize(a,[m1, n1]);

So while it may look like (the badly-named) a is a binary image, it's really a color image with just 2 colors: black and white. To fix, do this:

grayImage = rgb2gray(a);
binaryImage = grayImage > 128;

Then use binaryImage instead of a from that point on.

[Joshua Bone]

So imresize() doesn't support RGB images? Why does it have to be a binary image?

[Image Analyst]

imresize() works fine for RGB images as well as gray scale images, but you have to pass it the number of rows and columns. When she incorrectly did

% Never use size() like this with the array that comes directly out of imread()!
[m,n]=size(a); 

the (poorly-named) n was actually (the number of columns) * (the number of color channels), or if "a" were an RGB image, the number of color channels is 3 so n is 3 times as big as it should be.

See Steve's blog: http://blogs.mathworks.com/steve/2011/03/22/too-much-information-about-the-size-function/

Answer 2

[Tari]

Error using matlab.images.internal.resize.resizeParseInputs>parsePreMethodArgs (line 148)

When the second argument is a 1-by-3 vector, IMRESIZE interprets it as an M-by-3 colormap. If you intend to specify

an output size, use the syntax: imresize(A,___,'OutputSize',SZ) where SZ is a 1-by-2 vector.

[DGM]

While that's true, the error returned will be different than what OP was getting.

Error using matlab.images.internal.iptcheckmap (line 55)

Function IMRESIZE expected input number 2, MAP, to be a valid colormap. Valid colormaps cannot have values outside the range [0,1].

Obviously, that happens because any meaningful 1x3 size tuple in pixels is going to have values >1. Even if you gave it a nonsense sub-unity triplet, the error would still be

Error using imresize>fixupSizeAndScale (line 713)

Either scale or output size must be specified.

There are at least two problems with OP's code, if not three. @Image Analyst's link explains the problem with using size() carelessly, but due to what are likely other errors in OP's code, their misuse of size() probably didn't even get a chance to come into play.

Based on the error itself, we know that the image is RGB. This is supported by the fact that the file is a JPG. While single-channel JPGs exist, they are uncommon. Furthermore, there are no unit-scale JPGs (e.g. float or logical class). The fact that the geometry calculation is a function of the numeric scale of the data is probably causing a problem since OP seems to be presuming that the incoming image is in unit-scale.

% the inputs
a = imread('peppers.png'); % uint8, RGB (384x512x3)
b = 20000; % an unexplained magic number

% this is wrong, since it does not sum across pages
% or otherwise take them into account.
% it also doesn't really make sense that the pixel values
% (which are a strong function of the numeric class)
% should be used as a geometry scaling factor.
% so this might also be a conceptual bug and/or
% a misunderstanding of JPG functionality and numeric scale.
l = sum(sum(a))

l = 
l(:,:,1) =

    23746472


l(:,:,2) =

    13054194


l(:,:,3) =

    11331211

% this is also wrong, but n is never actually used, 
% which is probably also a bug
[m,n] = size(a)

m = 384

n = 1536

% this is a 1x1x3 vector 
m1 = fix(m*sqrt(b/l)) 

m1 = 
m1(:,:,1) =

    11


m1(:,:,2) =

    15


m1(:,:,3) =

    16

% since m is used instead of n for no apparent reason
% this is an exact duplicate of m1
n1 = fix(m*sqrt(b/l))

n1 = 
n1(:,:,1) =

    11


n1(:,:,2) =

    15


n1(:,:,3) =

    16

% the size argument is a 1x2x3 array
c = imresize(a,[m1 n1]); 

Error using matlab.images.internal.resize.resizeParseInputs>scaleOrSize (line 220)
Invalid scale or size input argument.

Error in matlab.images.internal.resize.resizeParseInputs>parsePreMethodArgs (line 163)
[scale, output_size] = scaleOrSize(next, next_arg);

Error in matlab.images.internal.resize.resizeParseInputs (line 28)
parsePreMethodArgs(varargin, method_arg_idx, first_param_string_idx);

Error in imresize (line 153)
params = matlab.images.internal.resize.resizeParseInputs(args{:});

This is my guess at what's intended. The user had a bunch of binary masks which they ruined by saving them as JPGs. Now they're RGB images, mis-scaled, and full of artifacts.

inpict = imread('i_ruin_my_images_with.jpg'); % this is a garbage JPG
outputarea = 20000; % a parameter

% clean up the input
inpict = im2gray(inpict);
inpict = imbinarize(inpict);

% rescale the image
inputarea = nnz(inpict); % get the total blob area
szi = size(inpict,1:2); % only get the specific dimensions needed
szo = fix(szi*sqrt(outputarea/inputarea)); % use vectors
outpict = imresize(inpict,szo);

Now the total blob area in the output image is as close as it can be to the specified parameter without exceeding it.