Is it possible to create an 'average' matrix from inputs of matrices of varying sizes?

The motivation for this question comes from one 3D matrix represents the coordinates of a cat walking through time.

Each matrix representing a cat takes on the form Matrix = [#points representing cat, x and y coordinate positions, #of frames]. The x and y coordinates comprising the cat is plotted succesively over each frame.

For example:

Cat 1(14, 2, 100)

Cat2(13, 2, 200)

Cat3(16, 2, 300)

My quesiton is whether there is a way to 'combine' all 3 cats so that I have one 'avg' cat? Maybe:

Cat_avg = (14, 2, 200) -> which is represented exclusively by the previous inputs?

Is such a thing possible? Would I have to augment smaller matricies w/ zeros or NaNs to do something like this? Apologies in advance if this is trivial; I have not yet taken linear algebra so am unsure if there is a simple way to do this, so am coming to this community for help. Thanks!

Answer 1

[yanqi liu]

sir，may be use reshape to vector，but its order is confuse，such as

clc; clear all; close all;
Cat1 = 1*ones(14, 2, 100);
Cat2 = 2*ones(13, 2, 200);
Cat3 = 3*ones(16, 2, 300);
Cat_all = cat(1, Cat1(:), Cat2(:), Cat3(:));

[Emmanuel Osikpa]

Hmm, I see what you mean with this; I think I can keep track of locations in my workspace and reshape with the corresponding coordinates. Thank you for your quick response! I will come back with update as to how this worked out