Undefined function 'dyaddown' for input arguments of type 'uint8'.

Getting the error:

Undefined function 'dyaddown' for input arguments of type 'uint8'.

Code is:

   %Clear command window.
clc;
    %Clear workspace.
clear;
    %Load the Lena file.
RGB = imread ('Lena.tiff'); 
    %Display the result of the conversion.
figure, imshow(RGB);
    %Convert RGB image to YCbCr Components.
YCbCr = rgb2ycbcr(RGB);
    %Isolate Y. 
Y = YCbCr(:,:,1);
    %Isolate Cb. 
Cb = YCbCr(:,:,2);
    %Isolate Cr. 
Cr= YCbCr(:,:,3);
    %downsample
dem = dyaddown(Cb,1,'m') % Downsample rows and columns
                        % with odd indices.

Why is this? How to I fix it?

Many thanks.

Answer 1

[Image Analyst]

I've never heard of dyaddown(). If you want to subsample Cb to get only the odd indexes and end up with an image 1/4 as big, do this:

dem = Cb(1:2:end, 1:2:end);  % Extract only odd indexes.

For even indices, do this:

dem = Cb(2:2:end, 2:2:end);  % Extract only even indexes.

[Alexander De-Ville]

Thank you very much, this works perfectly.

This is where I found dyaddown:

dyaddown Dyadic downsampling

There is also dyadup, but this does not work either for the same reason.

What would be the best way to upsample?

This exists:

<http://uk.mathworks.com/help/signal/ref/upsample.html>

But it only works in one dimension instead of x and y.

Thank you again!

[Image Analyst]

It's in the Wavelet Toolbox. Do you have that? Type ver on the command line to check.

[Alexander De-Ville]

Marked as accepted!

I have:

• MATLAB Version 7.14 (R2012a)
• Simulink Version 7.9 (R2012a)
• Control System Toolbox Version 9.3 (R2012a)
• Curve Fitting Toolbox Version 3.2.1 (R2012a)
• DSP System Toolbox Version 8.2 (R2012a)
• Image Processing Toolbox Version 8.0 (R2012a)
• Optimization Toolbox Version 6.2 (R2012a)
• Signal Processing Toolbox Version 6.17 (R2012a)
• Simulink Control Design Version 3.5 (R2012a)
• Statistics Toolbox Version 8.0 (R2012a)
• Symbolic Math Toolbox Version 5.8 (R2012a)

[Image Analyst]

So it looks like you don't have the wavelet toolbox so you can't run that particular function.

[Alexander De-Ville]

Silly me. Any hints on the upsampling issue?

What would be the best way to upsample?

This exists:

http://uk.mathworks.com/help/signal/ref/upsample.html

But it only works in one dimension instead of x and y. So cant undo 4:2:2 encoding.

Thank you again!

[Image Analyst]

You can use imresize() to specify the size of the output image you'd rather have.