Group neighboring cells with the same zoneID, MATLAB

Question

Say I have the following matrix of "zones"

brk_group =

     1     1     1     1
     3     2     2     3
     3     3     3     3

I would like to create a "grouping matrix" of these values. A group is defined as neighboring cells (top,right,bottom,left) with the same zone value.

A zone value can have multiple groups, since zones do not have to be joined spatially.

Data:

brk_group = [1 1 1 1; 3 2 2 3; 3 3 3 3]

What is wrong with "MY CURRENT CODE" below....

The zone value "3" in cell (2,4) is connected the the zone value "3" in cell (2,1) however when the loop in my current answer reached (2,4) is does not see this connectivity and will create a new group incorrectly dis-grouping cell (2,1) and (2,4)...

EDIT: Would like to just use the base MATLAB, no Image Processing Toolbox please! =)

CLARIFICATION: Zones are defined as ALL cells with the same zone value. Groups are defined as ALL spatially joined (top,bottom,left,right) cells with the same zone value.

MY CURRENT CODE:

function [groupMat groupIds] = createZoneGroups(zoneMat)
%
% groupMat = createZoneGroups(zoneMat)
%
% Groups zones spatially.
%
%   Input:
%       zoneMat: nxm matrix of "zones".
%
%   Output:
%       groupMat: nxm matrix of group ID's
%       groupIds: An array of unique group ID's
%
%   Note: A group is defined as spatially connected same zones. Diagonal
%       cells are not spatially connected.
%
%   Author: Kyle W. Purdon
%
groupIds = [];
groupMat = nan(size(zoneMat));
for rowIdx = 1:size(zoneMat,1)
    for colIdx = 1:size(zoneMat,2)

        % Check if the cell is nan, if so group(r,c)=nan
        if isnan(zoneMat(rowIdx,colIdx))
            continue;
        end

        % Check if the current cell has a group, if it does, break.
        if ~isnan(groupMat(rowIdx,colIdx))
            continue;
        end

        % Check the surrounding cells for groups, if any of the surrounding
        % cells (1) have a group, and (2) are in the same zone, assighn the
        % current cell that group. If not assign the current cell a new
        % group.

        % Check top cell
        if rowIdx > 1 && ~isnan(groupMat(rowIdx-1,colIdx))
            if isequal(zoneMat(rowIdx,colIdx),zoneMat(rowIdx-1,colIdx));
                groupMat(rowIdx,colIdx) = groupMat(rowIdx-1,colIdx);
                continue;
            end
        end
        % Check right cell
        if colIdx < size(zoneMat,2) && ~isnan(groupMat(rowIdx,colIdx+1))
            if isequal(zoneMat(rowIdx,colIdx),zoneMat(rowIdx,colIdx+1));
                groupMat(rowIdx,colIdx) = groupMat(rowIdx,colIdx+1);
                continue;
            end
        end
        % Check bottom cell
        if rowIdx < size(zoneMat,1) && ~isnan(groupMat(rowIdx+1,colIdx))
            if isequal(zoneMat(rowIdx,colIdx),zoneMat(rowIdx+1,colIdx));
                groupMat(rowIdx,colIdx) = groupMat(rowIdx+1,colIdx);
                continue;
            end
        end
        % Check left cell
        if colIdx > 1 && ~isnan(groupMat(rowIdx,colIdx-1))
            if isequal(zoneMat(rowIdx,colIdx),zoneMat(rowIdx,colIdx-1));
                groupMat(rowIdx,colIdx) = groupMat(rowIdx,colIdx-1);
                continue;
            end
        end

        % If the loop gets to this point, assign a new groupId to the cell.
        if isempty(groupIds)
            groupIds = 1;   % Start with group #1
        else
            groupIds(end+1) = groupIds(end)+1;  %Increment the last group by 1.
        end

        groupMat(rowIdx,colIdx) = groupIds(end);

    end
end
end

Answer 1

Solved by a code I had to really dig for on the MATLAB exchange.

It is essentially a "generalized" version of bwlabel() that uses the "Union-Find" algorithm.

http://www.mathworks.com/matlabcentral/fileexchange/26946-label-connected-components-in-2-d-array