I have a 3D point cloud representing ad object. I use a 3D cylinder to fit this object in the point cloud, so I check if each point is inside the cylinder and, if it is, then I assign a weight to that point.
So, at the end I have an estimate of how many points fall inside the cylinder and what is the weight of the fitting.
Now I would like to normalize the fitting dividing for the cylinder volume but this is in decimeter so the proportion I think it is not right. I think I can't count and weight points and divide them for a metric number. I think that I have to transform the metric volume into the "number of points that the cylinder can contain". This is just an idea of mine.
But how can I do so? Do you have any other idea?
The volume of the cylinder is $V=\pi r^2 h$ in any unit system as long as you measure $r,h$ in the same units of length and $V$ in units of that same length$^3$ I don't understand what you mean about weighting the points. Looking at your title involving pixels, if each point represents a certain volume, you can divide this volume into the cylinder volume to see how many points will fit. It will not be exact as the edge of the cylinder is round. The error will get smaller as the number of pixels increases.