Adaptive Savitzky-Golay filter

Hi

This code in matlab exchange adaptive SG

uses adaptive SG filter on a generated data and noise .

I need to apply this method on some signals , what should be changed in order to use it for my case ?

global Pkmi_array

m=(length(data_frame)-1)/2;
N=length(data_frame);
x=-m:m;
y=data_frame;
x=x(:);
y=y(:);

k_start=2;% Savitzky-Golay differentiation filters with polynomial degrees 1 and 2 are the same. 
k_end=N-1-2; 
alpha=0.01;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

 if k_end<=k_start 
    polynomial_degree=k_start;
    return;
else

    k=k_start;
    p_fit=polyfit(x,y,k);%polynomial degree = k_start
    res_even=sum((y-polyval(p_fit,x)).^2);

    while k<k_end %%%%%%%%%%%%%%%%k=2,4,...(even)
        %k

        n1=k;

        %polynomial degree = k
        k=k+1;
        value1=0;
        value2=0;
        for i=-m:m
            value1=value1+Pkmi_array(k+2,m,i+m+1)*y(i+m+1);
            value2=value2+(Pkmi_array(k+2,m,i+m+1))^2;
        end
        res_tmp=res_even(end)-value1^2/value2;

        %polynomial degree = k+1
        k=k+1;
        value1=0;
        value2=0;
        for i=-m:m
            value1=value1+Pkmi_array(k+2,m,i+m+1)*y(i+m+1);
            value2=value2+(Pkmi_array(k+2,m,i+m+1))^2;
        end
        res_even=[res_even res_tmp-value1^2/value2];

        n2=k;    
        v1=n2-n1;
        v2=N-n2-1;   
        Fx=(res_even(end-1)-res_even(end))/(n2-n1)/(res_even(end)+eps)*(N-n2-1);
        F_alpha=finv(1-alpha,v1,v2);
        % finv is a function in the Statistics Toolbox        
        % Inverse of the F cumulative distribution function

        if Fx<F_alpha
            break;
        end
    end    

    if Fx<F_alpha
        polynomial_degree=k-2;
    else
        polynomial_degree=k_end;
    end
end

[Jan]

@sarmad m: Is this the unchanged code from FileExchange: adpf_diff? Then please delete it or append the license text. See Answers: republish code from FileExchange .

 what should be changed in order to use it for my case

How could the readers know this? What is the difference between "your case" and the standard input for this function?

Answer 1

[Image Analyst]

Put some code at the top of it to define, or read in, your x and y data.

[sarmad m]

My x data is the frames from 1:1080 and y is signal values . I have attached it . I have tried to read x and y data but I can't get it work as the demo in the example. Could you please help me

[Image Analyst]

Make it EASY for me to help you. csvread() did not work with your data and it's all in one column anyway so I don't know what's x and what's y. Please give me the code to read in your data file and create x and y. If you wait for me to do it, it might not be until tomorrow or later, if ever.

http://www.mathworks.com/matlabcentral/answers/6200-tutorial-how-to-ask-a-question-on-answers-and-get-a-fast-answer

[sarmad m]

The signal like shown below

Y values are in attachments. To read Y I just do normal import and plot(Y) to show the signal .