How to fix error in port widths or dimentions on the reshape block input?
8 views (last 30 days)
Alexander Semenov on 16 May 2021
I need to make DCT image compression in Simulink. Image size by default has 3 dimensions, so I used reshape tool to remove the third dimension. Color Space Conversion tool only accepts 3 dimensional images, so I used reshape again. When I try to run the simulation, I get two following errors:
Error in port widths or dimensions. 'Input Port 1' of 'untitled/Reshape1' has 786432 elements. This port does not accept the dimensions (or orientation) specified by the input signal.
I thought reshape was supposed to accept any matrix as an input.
Andy Bartlett on 18 May 2021
The purpose of Simulink's Reshape block and MATLAB's reshape command is only to "reinterpret" the dimensions (aka size) of the signal/variable. The number of elements is not changed. The values as they would appear in the "flat memory" of the computer would also be unchanged.
Your usage doesn't fit the reshape behavior. Your input to the first reshape block has this many elements.
inputNumel = prod([512,512,3])
Your desired output has this 1/3 as many elements.
outputNumel = prod([512,512])
Since you are trying to change then number of elements, the reshape block is throwing an error.
One approach to solve this is to put the following code into a MATLAB Function block
function [Y,Cb,Cr] = splitData(u)
Y = u(:,:,1);
Cb = u(:,:,2);
Cf = u(:,:,3);
But there is probably a simpler way to achieve your high level goal using capabilities of Image Processing Toolbox.