you can perform what you're doing a bit simpler.
E = randi(100,6000,3); %my generation of dummy data
windowL =500;
thresh= 50;
for ind = 1:length(E(:,1))-windowL+1 % set the end such that you don't index more than the length of the array E.
Window = E(ind:ind+windowL-1); %pulled out section of E to be used
tWindow = Window>thresh; % threshold the window you'll get a 500x3 logical array
stWindow = sum(tWindow); % then take the sum to count how many are over the threshold per column (you'll end up with a 1x3 matrix of how many are over the threshold.
tstwindow = stWindow>windowL/2; % then you can threshold again to see which ones have a majority over the threshold.
detect(ind,:) = ttwindow; %then set your detected array based on the last calculation.
end
I broke up the code a bit to show some intermediate steps showing how using logical operations you can perform these things.
