[RESOLVED] Separating list of points into different groups

In the program I am working on, we analyze the data of MRI scans. Sometimes, we want to draw boundaries around certain vessels in the brain.

When creating an object boundary around two different vessels, it is normal to use two different objects.

However, sometimes the second boundary is created under the same object name. This means that there are two separate boundaries that are technically just one object.

I am trying to make a function that removes a region from one object, because currently the only option is to delete the object as a whole.

The boundary is stored as a CList of CPoints, and these points only contain the actual boundary, not the points within the boundary. Here is an example output of the boundary points relative to the mouse:

x: 269 y: 217 mouse: 234 217
x: 260 y: 215 mouse: 234 217
x: 261 y: 214 mouse: 234 217
x: 262 y: 213 mouse: 234 217
x: 263 y: 212 mouse: 234 217
x: 265 y: 212 mouse: 234 217
x: 266 y: 213 mouse: 234 217
x: 267 y: 214 mouse: 234 217
x: 268 y: 215 mouse: 234 217
x: 260 y: 214 mouse: 234 217
x: 261 y: 213 mouse: 234 217
x: 262 y: 212 mouse: 234 217
x: 266 y: 212 mouse: 234 217
x: 267 y: 213 mouse: 234 217
x: 230 y: 213 mouse: 234 217
x: 229 y: 212 mouse: 234 217
x: 229 y: 213 mouse: 234 217
x: 229 y: 214 mouse: 234 217
x: 230 y: 212 mouse: 234 217
x: 230 y: 214 mouse: 234 217
x: 231 y: 212 mouse: 234 217
x: 231 y: 213 mouse: 234 217
x: 231 y: 214 mouse: 234 217
x: 230 y: 211 mouse: 234 217
x: 232 y: 213 mouse: 234 217
x: 229 y: 211 mouse: 234 217

As you can see, about half way through, the points in the X column change drastically from 267 to 230. So, with a difference of over 35, it is obvious where the two regions could be separated.

However, it is not always that easy. If the two regions are extremely close together, then the differences between the X and Y points would be almost impossible to differentiate between one region and another.

As you can see at the top of the list of points, the X variable changes by a factor of 9 units, even though it is still a part of the same region. So if I had a second region less than 9 units away, then it would be seemingly impossible to tell the difference.

Any ideas?

I think the answer is obvious, and you might not like it: if there is nothing in your list of CPoint’s that tells you where one object ends, you can’t differentiate between objects simply looking at this list.
When you draw those boundaries, could you somehow mark when mouse’s button was down and when up?

Quote:

---

Originally Posted by VladimirF

I think the answer is obvious, and you might not like it: if there is nothing in your list of CPoint’s that tells you where one object ends, you can’t differentiate between objects simply looking at this list.
When you draw those boundaries, could you somehow mark when mouse’s button was down and when up?

---

Well, this boundary drawing is done by "region growing" and just takes one mouse click, but thank you for saying what you did because I think I understand where you are going with that.

The object clist is actually comprised of CPoint4d objects, and I just noticed that the cpoint4d.v variable was not being used in any of the algorithm, so in the Region Grow function, when it creates the final CPoint to be added into the list, I set cpoint4d.v to a unique float variable, thus signifying the end of the region!

Thank you so much though, because I would not have thought of this without your post

Quote:

---

Originally Posted by dchaacke

Thank you so much though, because I would not have thought of this without your post

---

You are very welcome!