Find consecutive numbers below threshold using bwlabel

Hi,

I want to find all consecutive values and their region in a vector that are below a certain threshold.

I am able to do this using the following code. In this example I find 5 regions below the set threshold which is correct.

y = [1 2 1 2 3 4 5 4 3 2 0.4 0.3 0.4 0.4 0.6 0.4 0.5 0.4 0.6 0.5 0.3 0.5 0.6 0.2 1 2 3 4 1 2 3 2 3 2 5 4]
plot(y)

lowThreshold = 0.5

% label points below threshold as 1
[L, count] = bwlabel(y < lowThreshold); % label waves below threshold
% Get lengths of each region
props = regionprops(L, 'Area', 'PixelList');
regionLengths = [props.Area];
for k = 1 : count
    fprintf('Region #%d is %d elements long and starts at element #%d\n',...
        k, regionLengths(k), props(k).PixelList(1,1));
end

Now I want to modify this code to count the 5 regions as 1 connected big region although I exceed my threshold by a tiny amount. How can I do this? Take first index of region 1 and last index of region 5 and count as 1 large region. For example:

Thanks.

Answer 1

[Steve Eddins]

You could a hysteresis thresholding technique. First, threshold with a strict threshold, like your 0.5. Then, threshold with a slightly looser threshold, perhaps 0.1. Next, allow the strict threshold result to grow into adjacent regions that are included in the looser threshold result. The Image Processing Toolbox function imreconstruct can be used to do this. Here's a modified version of your code:

y = [1 2 1 2 3 4 5 4 3 2 0.4 0.3 0.4 0.4 0.6 0.4 0.5 0.4 ...
    0.6 0.5 0.3 0.5 0.6 0.2 1 2 3 4 1 2 3 2 3 2 5 4];
plot(y)
lowThreshold = 0.5;
medThreshold = 0.7;

mask1 = y < lowThreshold;
mask2 = y < medThreshold;
combined_mask = imreconstruct(mask1,mask2);

% Get lengths of each region.
% Note that regionprops can take a binary mask directly as input.
props = regionprops(combined_mask, 'Area', 'PixelList');
regionLengths = [props.Area];
for k = 1 : numel(props)
    fprintf('Region #%d is %d elements long and starts at element #%d\n',...
        k, regionLengths(k), props(k).PixelList(1,1));
    hold on
   [r1,r2] = bounds(props(k).PixelList(:,1));
   plot([r1 r2],[0.5 0.5],'Color','r','Linewidth',3)
   hold off
end

Region #1 is 14 elements long and starts at element #11

[Konvictus177]

Wow, exactly what I was looking for. Love the solution so far.

I will test it a bit more and come back in case of any issues.

Thank you!