How to deal with this error ??

Error using + Matrix dimensions must agree. *Error in (line 77) y= d1w + d2w + aa2;

d1 and d2 are details coefficients at level 1 and 2 respectively. aa2 is the approximate coefficient at level 2

ech is an ecg signal

 %wavelet decomposition  
    [C, L] = wavedec(ech, 2,'haar');   

 %extraction of approximate and detail coefficients of signal ech
    aa2 = appcoef(C, L, 'haar', 2);
    dd2 = detcoef(C, L, 2);
    dd1=  detcoef(C, L, 1);
%Wiener computations
s1=median(abs(d1))/.6745;
g1=((dd1).^2)/((dd1).^2)+((s1).^2);
d1w=repmat(d1(:), 1, size(d1)) .*g1;

 s2=median(abs(d2))/.6745;
g2=((dd2).^2)/((dd2).^2)+((s2).^2);
d2w=repmat(d2(:), 1, size(d2)) .*g2;
%reconstruct the signal
subplot(3,2,5);
y= d1w + d2w + aa2;
plot(y,'k');

[Walter Roberson]

"ech is an ecg signal"

ecg signals are multiple channel, typically either 5 or 12. The code you are using requires that ech is a vector.

Your d1 is not defined in the code. Did you possibly intend to code dd1 instead of d1 ?

[Good mind]

    clear all;
    subplot(3,2,1);
    load ecg.m;
     plot(ecg); title ('ECG');
    xlabel('samples'); ylabel('magnitude');

    %wavelet decomposition 
    [C1, L1] = wavedec(ecg, 2,'haar');

    %extraction of approximate and detail coefficients of signal
    a2 = appcoef(C1, L1, 'haar', 2);
    d2=  detcoef(C1, L1,2);
    d1=  detcoef(C1, L1, 1); 

    % hard (h) thresholding
     TM=1;
    n2=length(d2);
    s2=median(abs(d2))/.6745;
    t2= TM*s2;  
    d2thard = wthresh(d2,'h',t2);

    n1=length(d1);
    s1=median(abs(d1))/.6745;
    t1= TM*s1;   
    d1thard = wthresh(d1,'h',t1);  

    %wavelet coefficients after hard thresholding 
    hard_ua2 = upcoef('a',a2,'haar',2, 2000);
    hard_ud2 = upcoef('d',d2thard,'haar',2, 2000);
    hard_ud1 = upcoef('d',d1thard,'haar',1, 2000);

    subplot(3,2,3);
    ech= (hard_ud1 + hard_ud2 + hard_ua2); 
    plot(ech);title('ech the signal after hard thre');

%wavelet decomposition 2 
  [C, L] = wavedec(ech, 2,'haar');   

 %extraction of approximate and detail coefficients of signal ech
    aa2 = appcoef(C, L, 'haar', 2);
    dd2 = detcoef(C, L, 2);
    dd1=  detcoef(C, L, 1);

    %Wiener computations
g1=((dd1).^2)/((dd1).^2)+((s1).^2);
d1w=repmat(d1(:), 1, size(d1)) .*g1;

 g2=((dd2).^2)/((dd2).^2)+((s2).^2);
 d2w=repmat(d2(:), 1, size(d2)) .*g2;

 %reconstruct the signal
subplot(3,2,5);
y= d1w + d2w + a2;
plot(y,'k');

Answer 1

[Walter Roberson]

You have

 d1w=repmat(d1(:), 1, size(d1)) .*g1;

 g2=((dd2).^2)/((dd2).^2)+((s2).^2);
 d2w=repmat(d2(:), 1, size(d2)) .*g2;

size(d1) is a vector, so your repmat is working as if you had used repmat(d1(:), [1 1 500]) when you wanted repmat(d1(:), [1 500]) I guess. Same problem with your repmat for d2w. (Ah, I just realized that 500 is length(ecg)/2 )

When you get down to

y= d1w + d2w + a2

you run into the fundamental problem that arrays derived from d1 are length(x)/2 on a side, but arrays derived from d2 are length(x)/4 on a side. You will not be able to add them.

[Good mind]

thank you sir. i don't understand even when i use d1w= d1 .*g1 also i get *Error using . Matrix dimensions must agree

 %Wiener computations
g1=((dd1).^2)/((dd1).^2)+((s1).^2);
d1w= d1 .*g1;

 g2=((dd2).^2)/((dd2).^2)+((s2).^2);
d2w=d2 .*g2;

the length of ecg is 2000

[Walter Roberson]

change

 d1w=repmat(d1(:), 1, size(d1)) .*g1;

 g2=((dd2).^2)/((dd2).^2)+((s2).^2);
 d2w=repmat(d2(:), 1, size(d2)) .*g2;

to

 d1w=repmat(d1(:), 1, length(d1)) .*g1;

 g2=((dd2).^2)/((dd2).^2)+((s2).^2);
 d2w=repmat(d2(:), 1, length(d2)) .*g2;

That will get you through to

y= d1w + d2w + a2;

where you will have problems because d1w is 1000 x 1000, d2w is 500 x 500, and a2 is 500 x 1 . You are attempting to add matrices that are inherently differently sizes, one half the size of the other. You need to go back and figure out what you are trying to do there.

[Good mind]

i used it (length instead size) before, but i got the same problem Error using + Matrix dimensions must agree.

Error in y= d1w + d2w+a2;

[Walter Roberson]

You are attempting to add matrices that are inherently differently sizes, one half the size of the other. You need to go back and figure out what you are trying to do there.