How can I group adjacent not-empty cells?

Question

I have a 4x4 cell array C, which

C= {

[1] [3] [6] [ ]; 

[2] [ ] [ ] [8];  

[ ] [4] [ ] [9]; 

[ ] [5] [7] [ ]}

I want to generate a new cell array D which give me

D = {[1;2], [3], [4;5],[6],[7],[8;9]}

basically I want to 1. combine the adjacent non empty cell in each column vertically and 2. output the new cell array D contains the result.

Answer 1

You can use this. I've used bwlabel from the imaging toolkit:

C= {                   ...
[1] [3] [6] [ ];       ...
[2] [ ] [ ] [8];       ...
[ ] [4] [ ] [9];       ...
[ ] [5] [7] [ ]};

lenf = @(X)~isempty(X);
lens = cellfun(lenf, C);

lens is now a logical array indicating if any slot in C is empty or not. Now we can construct D by treating each column in lens as a 1 x whatever binary image, and seek regions using bwlabel(). Finally we put the regions into D.

sum = 0;
for k = 1:size(lens,2)
    [L,num] = bwlabel(lens(:,k), 4);
    for idx = 1:num
        D{idx+sum} = cat(1, C{L==idx, k});
    end
    sum = sum + num;
end