How can I compute the mean and variance of a channel using blockproc?

I amy trying to compute the block-based mean and variance of each channel of an image using blockproc through block of size 8x8. After splitting the image into its channels (in the RGGB format), I used blockproc as for the red channel (I_r)

fun1 = @(block_struct) mean(block_struct.data);
mean_ch = blockproc(I_r, [8 8], fun1);
fun2 = @(block_struct) var(block_struct.data);
variance = blockproc(I_r, [8 8], fun2);
scatter(mean_ch, variance); %scatter plot of mean and variance 

I get the error saying that blockproc has too many input arguments. Considering I took the syntax from the documentation, I am unable to figure what the problem is. Why does it say "too many input arguments"? Is there a different way to find mean and variance with blockproc? Any help is appreciated.

[Walter Roberson]

which -all blockproc

/MATLAB/toolbox/images/images/blockproc.m

You should see one output that looks similar to the above, but in the directory that you installed MATLAB in to.

I suspect you will see two files instead, and that the first one is a third-party file that is interfering with your use of the Mathworks code.

[Deepika Sundresh]

I just saw one entry like the one you have written. There wasn't a dummy file.

Answer 1

[Image Analyst]

You need to use (:) to turn the block data into a column vector, otherwise you get multiple values instead of one.

%===============================================================================================================================
% Define the function that we will apply to each block.
% First in this demo we will take the standard deviation gray value in the block
% and create an equal size block where all pixels have the standard deviation value.
% Image will be the same size since we are using ones() and so for each block
% there will be a block of 8 by 8 output pixels.
stdFilterFunction = @(theBlockStructure) std(theBlockStructure.data(:)) * ones(size(theBlockStructure.data), class(theBlockStructure.data));
% Note: it's the ones(size(theBlockStructure.data), class(theBlockStructure.data)) that 
% makes the output be the same size as the input rather than 1/8th the size of the input.  It basically replicates the standard deviation pixel into an 8x8 square.
%===============================================================================================================================
% Standard Deviation of 8 pixel by 8 pixel block
% Block process the image to replace every pixel in the 
% 8 pixel by 8 pixel block by a single pixel whose value is the standard deviation of the pixels in the block.
blockSize = [8, 8];
% Quirk: must cast grayImage to single or double for it to work with std().
% blockyImage8888 = blockproc(grayImage, blockSize, stdFilterFunction); % Doesn't work.
blockyImage8888 = blockproc(double(grayImage), blockSize, stdFilterFunction); % Works.
[rows, columns] = size(blockyImage8888);

See my attached demos for a full demo.

Or you could try mean2() instead of mean(), and var2() instead of var().

[Deepika Sundresh]

Thank you. I tried mean and mean2 both, and I got the same error. I then ran your demo code for mean, and I got this error:

Error using blockproc

Too many input arguments.

Error in blockproc_mean (line 54)

blockyImage8888 = blockproc(grayImage, blockSize, meanFilterFunction); % Works

Now I am unable to understand why this is happening. I tried with a different image as well.

[Image Analyst]

What does this say

>> which -all blockproc

You didn't, by chance, name your script "blockproc.m" did you? Never name your scripts or functions after built-in functions.

[Deepika Sundresh]

I got this output:

>> which -all blockproc

/matlab/toolbox/images/images/blockproc.m

I haven't named my script blockproc.m :)

Answer 2

[Matt J]

It will be more efficient if you download sepblockfun() instead,

https://www.mathworks.com/matlabcentral/fileexchange/48089-separable-block-wise-operations

A=double(rgbImage);

Means=sepblockfun(A,[8,8,1],'mean');

Vars=sepblockfun(A.^2,[8,8,1],'mean') - Means.^2;

[Deepika Sundresh]

The requirement for me is to use blockproc only, and that is why i am facing issues. Could you tell me how I can do it blockproc itself?

Answer 3

[Deepika Sundresh]

I was able to execute the code properly, it was that an older function was saved as blockproc when I downloaded the assignment, and it was interfering with my code. I deleted it and the blockproc works fine. Thank you all for the help!