How to create for loop to find coefficients of segments?

Hi all. I have a data matrix named: y (150x240)

I'm using each column (150 datas) of that matrix to find some parameters.

For example, for first frame:

y1=y(1:150)
p=10;
[a,g] = lpc(y1,p)

And same for second column(y2=y(151:300)) I need to find [a,g] for all columns. How can I create for loop for this?

[Star Strider]

Linear prediction filter coefficients: lpc

Answer 1

[Image Analyst]

Try this:

[rows, columns] = size(y);
p=10;
for col = 1 : columns
    thisColumn = y(:, col);
    [a(col), g(col)] = lpc(thisColumn, p);
end

[jack star]

I got that error: Subscripted assignment dimension mismatch.

I guess it is because a is matrix (1x11) and g is single number. For example first frame results manually:

[Image Analyst]

Looks like it's because you tried your original code and totally ignored the code I suggested to you.

I have no idea what the lpc function does but I assume it returns two scalars.

[jack star]

[a(col), g(col)] = lpc(thisColumn, p);

when I delete the parenthesis of a(col) and g(col), it works, but it finds only last results of a and g.

How can we find a1,g1 , a2,g2 , ...

[Image Analyst]

OK - it turns out that lpc is a function in the Signal Processing Toolbox. I don't use that function or know what it does. But if you have just [a,g] then you're overwriting a and g on every iteration so of course it will have only the values from the last iteration. That's why you need to index them. If g is a scalar and a is a 11 element array, you can do

[a(col, :), g(col)] =  lpc(thisColumn, p);

Before the loop, preallocate a:

a = zeros(columns, 11);

[jack star]

Thank you very much for your help.