How to smoothen a plot?

a=imread('IMG_0.jpg');

b=imread('IMG_20.jpg');

c=a(:,:,3)-b(:,:,3);

max(max(c));

min(min(c));

no_row=size(c,1);

no_col=size(c,2);

k=zeros(1,4857);

y=linspace(0,200,200);

l=zeros(45,200);

for m=1:45

for n=1:200

l(m,n)=mean(mean(c(((m-1)*10+1):10*m,((n-1)*24+1):24*n)));

end

end

o=zeros(1,200);

for i=1:200

p=l(:,i);

for j=1:45

if p(j)>=100

o(i)=j;

plot(y,(45-o)*20/45)

break

end

end

end

how do I smoothen the plot and get rid of the vertical lines?

[DARSHAN KUMAR BISWAS]

Please keep in ming that I am a begineer and before matlab I have done work only in C loanguage.

[Image Analyst]

Yes, most people here are beginners. But that doesn't mean you cannot understand a direct request to attach your data? Anyone can understand that even if they don't know MATLAB programming. Just use the paperclip icon to attach a text file with the x and y data in it. Not sure what those two images you read in are, but go ahead and attach those two images just in case it helps explain something.

Answer 1

[Image Analyst]

Try this (untested because you forgot to attach your data). Also it depends if you want to delete the bad points or it you want to change them to interpolated/repaired values.

% To remove bad elements:
badIndexes = y > y(end);
x(badIndexes) = [];
y(badIndexes) = [];

% Alternatively to leave them there but replace them with a quadratic fit:
goodIndexes = (y <= y(end)); % A logical vector.
% Now fit a quadratic to just the good values.
coefficients = polyfit(x(goodIndexes), y(goodIndexes), 2);
% Now get a fitted y.
yFixed = polyval(coefficients, x);
% Replace bad y values with fixed/repaired/interpolated ones.
y(~goodIndexes) = yFixed(~goodIndexes); 

If you have any more questions, then attach your data and code to read it in with the paperclip icon after you read this:

TUTORIAL: How to ask a question (on Answers) and get a fast answer

Answer 2

[John D'Errico]

Simple. Delete the points in the beginning of the curve that give you the spikes. WTP? They will be easy enough to spot, as they will have an unreasonably large y value compared to the rest.

If the result is STILL too noisy, then use smooth on it. Or appy a smoothing spline, if you hve the curve fitting toolbox. Or, to be honest, the remaining curve looks almost well enough behaved that you could fit it with a low order polynomial. You can do that using polyfit, or with the stats toolbox or the curve fitting toolbox. You might prefer the results from a robustfit, which will be than from a simple least squares fit, given the somewhat spiky data.

If you want a better answer, then you would need to attach a copy of your data to a comment, or by editting your original question.