hi

1.    float sum = 0;
2.    int n = 0;
3.    float averageA = 0;
4. 
   
5.    for (int ix=0;ix<gw;ix++) {
6.         int x0=ix*gs+gs2;
7.         for (int iy=0;iy<gh;iy++) {
8.           int y0=iy*gs+gs2;
9.           int ig=iy*gw+ix;
10. 
    
11.           float u=df*sflowx[ig];
12.           float v=df*sflowy[ig];
13. 
    
14.           // draw the line segments for optical flow
15.           float a=sqrt(u*u+v*v);
16. 
    
17.           //Average a ? 
    
18.           n++;
19.           sum += a;
20.            averageA = sum / n;
    
21.        }
22.     }

Thank you very much! This works great!

If you are only interested in non-zero values then

1.    float sum = 0;
2.    int n = 0;
3.    float averageA = 0;
4.     for (int ix=0;ix<gw;ix++) {
5.         int x0=ix*gs+gs2;
6.         for (int iy=0;iy<gh;iy++) {
7.           int y0=iy*gs+gs2;
8.           int ig=iy*gw+ix;
9.           float u=df*sflowx[ig];
10.           float v=df*sflowy[ig];
11.           // draw the line segments for optical flow
12.           float a=sqrt(u*u+v*v);
13.           //Average a ? 
14.           if(a != 0){   // only consider non-zero values
    
15.               n++;
16.               sum += a;
17.               averageA = sum / n;
18.           }
19.        }
20.     }

If you are looking at a stream of data to see if if is trending towards a particular value then you need to calculate the
moving average. For more info on that look at this topic

Mushussu thank you so much for your suggestion. So far it works, I have to do some more tests to see if it fits well my current application