smoothing filter, how to avoid initial zero value

This is my problem: I have a vector that represents a trajectory. the first value is 420 and it decreases ti 150. this trajectory is noisiy so I filtered it using the "filter" function. the code is:

b=ones(1,10)/10; zfil = filter(b,1,z);

but when I plot zfil it starts from a zero value. How could I avoid this problem?

theese are the plots , the first is z, the second zfil

https://www.dropbox.com/s/k410rcpazz777gq/untitled.jpg

thanks!

Answer 1

[Wayne King]

One thing you can do is to compensate for the delay introduced by the smoothing filter. I'll create a noise signal x to illustrate, but substitute your signal for x.

    b = ones(10,1)/10;
    D = mean(grpdelay(b,1));
    % let x be your input signal
    x = randn(100,1);
    % append some zeros to account for the delay
    x = [x; zeros(round(D),1)];
    y = filter(b,1,x);
    y = y(round(D)+1:end);

Another thing you can try is to use filtfilt() but that squares the magnitude response of the filter

      y = filtfilt(b,1,x);