## How to effectively reorder data

### Darlington Mensah (view profile)

on 23 Nov 2018
Latest activity Commented on by Image Analyst

on 23 Nov 2018

### Bruno Luong (view profile)

I have unordered coordinates and when i order it by doing th following,
contx = table2array(contorno(:,1));
conty = table2array(contorno(:,2));
cx = mean(contx);
cy = mean(conty);
a = atan2(conty - cy, contx - cx);
[~, order] = sort(a);
reorderedx = contx(order);
reorderedy = conty(order);
When I plot, this is what i get
plot(reorderedx, reorderedy)
How do I get a continuous "smoth" line plot without the spikes that are present in the figure.
What is the best way to reorder my data?

### Bruno Luong (view profile)

on 23 Nov 2018

Solve it as Traveling SalesMan PB: Use this FEX
contx = table2array(contorno(:,1));
conty = table2array(contorno(:,2));
userConfig = struct('xy',[contx(:) conty(:)]);
resultStruct = tsp_nn(userConfig);

### Cris LaPierre (view profile)

on 23 Nov 2018

This is a non-trivial problem, as I'm sure you've discovered. I'm wondering if you can find something useful in this traveling salesman example problem.

### Image Analyst (view profile)

on 23 Nov 2018

One way is to sort things angulalry. Compute the center, then compute the angles of each point from the center, then use both outputs of sort(). Easy, but let us know if you can't do it.

Show 1 older comment
Image Analyst

### Image Analyst (view profile)

on 23 Nov 2018
Honestly I didn't look at the code much. The plot does not look like the points are all going in the same angular direction. There is a lot of backtracking going on.
Bruno Luong

### Bruno Luong (view profile)

on 23 Nov 2018
Of course it does, depending where the center is (I can see very well where it's located approximatively, up north there), which is in turn depend on the density of the point on the boundary.
In anycase this kind of method (angle sorting) bounds to fail for such complicated boundary.
Image Analyst

### Image Analyst (view profile)

on 23 Nov 2018
Yes, when I ran the code, and plotted the centroid, it shows the centroid is not where you'd have thought at first. For what it's worth, I also tried the boundary() function. I agree the TSP solution seems to be best. I'm voting for that.