## strange output after using the Moving average filter

### Gova ReDDy (view profile)

on 26 Dec 2013
Latest activity Commented on by Image Analyst

### Image Analyst (view profile)

on 26 Dec 2013

Hello, I used a moving average filter of 30 order as shown below for the attached signal and the output of the moving average filter starts from the some other value instead of starting from the original input value when it comes for filtering the new set of values there is new set input values as shown in the image with the green portion

```P=load('Noise_Pulse_signal.mat');
a2=P.a1;
x1=a2(1:500,:);
```
```   for i=1:500:length(a2)-1500
x1=a2(i:i+499,:);
movingAverage = conv2(x1,ones(30,1)/30, 'same');
Mov_avrg(i:i+499,:)=movingAverage;
input(i:i+499,:)=x1;```
```   end;
plot(input);hold on;plot(Mov_avrg,'-r');
hleg1 = legend('input','MovAvrg');```

Can someone explain the reason for this.

### Image Analyst (view profile)

on 26 Dec 2013

The 'same' option assumes that the signal is zero outside of the signal. So the part of the window that spills outside the signal is zero, which is less than your signal. So it makes absolute prefect sense that the first 15 and last 15 elements of your moving average signal will be lower than the input signal.

Gova ReDDy

### Gova ReDDy (view profile)

on 26 Dec 2013

thankyou and Is there any way of modifying the filter so that the filter will response will start from the same intial value as the input.

Image Analyst

### Image Analyst (view profile)

on 26 Dec 2013

I don't think you know how the filter works. To have the window fully covering the signal, the center of the window has to be half a window's width inside the signal. Just draw it out and you'll see. There is a 'valid' option which will use that case but the output will be a window's width shorter than the input signal for obvious reasons.