erosion and dilation in signal processing

we need codes for erosion and dilation in signal processing...as imdilate and imerode functions can only be used in image processing..

Answer 1

[Image Analyst]

They can be used for 1D signals, if you have the Image Processing Toolbox. Do you have that toolbox? If not, you might play around with this approach:

windowSize = 3;
originalSignal = rand(10,1)
hank = hankel(originalSignal)
h3 = hank(:,1:windowSize)
dilatedSignal = max(h3, [], 2)

[Annapoorani]

no..can we do it for ECG signal?

[Image Analyst]

Why do you say "no"? Why shouldn't you be able to do it for an ECG signal. The procedures don't care what the signal is - they're all just a bunch of arbitrary numbers as far as they're concerned.

[Walter Roberson]

I think the "no" might be for "Do you have that toolbox?"

[Image Analyst]

Oh, right, probably so. The hankel though is in base MATLAB and with a little modification of the code to handle edge effects it could do it. Or else use a for loop and the max() and min() functions to do it the intuitive, brute force way. Erosion is the local min and dilation is the local max.

[Annapoorani]

ya no is for the toolbox,sorry abt tat... mm thanks for answering,i wil try it and let yo know

[Annapoorani]

we tried imdilate got the answer but imgage is not displaying,it comin as image too large to fit on screen.so what should we do to view the dilated image

[Image Analyst]

If you don't have the Image Processing Toolbox, then how did you try imdilate()?

[John Cristion]

Kernel may need to be transposed

[Image Analyst]

You can still view the dilated image even though you get a warning that says it will shrink it down to fit onto the screen. You can safely ignore that warning. It's not serious. If you want to see it pixel for pixel, use the zoom tool to zoom in.

Answer 2

[shelley]

Yes, it is doable. It is better to binarize the signal first, and then use the command like

sel=strel('line',120,90);

%the length 120 depends on how big the gap you want to close/seal.

%the degree 90 or 0 depends on your 1D signal is in vector or row format;

results=imdilate(1Dsig,sel);