Sliding window: array gets smaller

I am currently working on implementing a sliding window into my code. This is is what i have got so far:

windowLength = 10;
        for i = 1:length(green)-windowLength
            greenDC(i) = mean(green(i:i+windowLength-1));
            redDC(i) = mean(red(i:i+windowLength-1));
            greenAC(i) = std(green(i:i+windowLength-1));
            redAC(i) = std(red(i:i+windowLength-1));
            
            %other codes
        end 

My problem is now, that i want to plot my results i get later in the code over the time axis t. But after my sliding window the arrays get smaller by 10 and now my time array is to big for the plotting to work.

Does anybody know how to solve this problem? Or is my sliding window completly wrong?

I already tried to interpolate the time, but its not working.

thanks in advance!

[Rik]

Is there a reason you want to avoid movmean or movstd?

[KALYAN ACHARJYA]

Have you checked imfilter() funtion to slide the window (effiecint way)?

[Adam Danz]

The number of elements in your smoothed array size should only differ by "windowLength". You can shorten the length of your time vector by the same number of units for plotting. Smoothed data at time t(n) represents data smoothed between the interval of t0(n) and t0(n)+windowLength-1.

[Milena]

@Rik I didnt get good results with movmean and movstd, so I decided to build a for loop and try it again

[Rik]

What exactly do you mean? Were the values you got unexpected? Or did you run into errors?

[Milena]

The error I calculated got bigger with movmean than without

[Rik]

And how did you determine that this was due to an incorrect implementation and not inherent to your data?

Answer 1

[Image Analyst]

If you want to shrink the window, try this (untested)

windowLength = 10;
for i = 1:length(green)
    index2 = min([length(green), i + windowLength - 1]);
    greenDC(i) = mean(green(i:index2));
    redDC(i) = mean(red(i:index2));
    greenAC(i) = std(green(i:index2));
    redAC(i) = std(red(i:index2));    
    %other codes
end 

You know, imfilter has edge effect options, including shrinking window as it approached the edge of the signal or image.

[DGM]

Are you sure about that? AFAIK, imfilter() handles all its edge treatment by padding. There are options for how the padding is generated and the extent of the returned array, but I don't recall anything about kernel truncation options. Same goes for medfilt2(), stdfilt(), etc.

EDIT: maybe you mean movmean(), movmedian(), etc. Those have truncation options.

[Image Analyst]

@DGM, you're right.

[Rik]

I believe the default behavior of this or a related function changed around R2017b. When I get home I will look up what function exactly and what the change was.

Edit: turns out it was R2017a, where imclose pads the image by half the size of the SE.