Timeline for How can I adaptively simplify a curved shape?
Current License: CC BY-SA 3.0
14 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jul 26, 2017 at 8:02 | comment | added | Szabolcs | @Mr.Wizard Beyond this one point that you mentioned, I do not have time to revise it right now. If you would like to do that, that would be great. I know that it is not very fast. | |
| Jul 26, 2017 at 7:51 | comment | added | Mr.Wizard | Do you have interest in revisiting and optimizing this code? If so I may wait for you to do that; if not I may try myself. | |
| Jul 26, 2017 at 7:43 | comment | added | Szabolcs |
@Mr.Wizard I do not remember ... I wrote the code years before I posted it here. I should have commented it better, but I did not. I may have run into tolerance problems. As I remember there are cases when a==b is True but DeleteDuplicates[{a,b}] does not remove one of the elements because there is a tiny difference between them.
|
|
| Jul 26, 2017 at 7:30 | comment | added | Mr.Wizard |
The pairwise compare of DeleteDuplicates[points, #1 == #2 &] is very expensive as known. What is its purpose here?
|
|
| Feb 22, 2017 at 9:19 | comment | added | Alexey Popkov |
Finally I come to dividing the original shape into two parts but for another reason: the marker should be (anti)symmetric. With SimplifyCurve I got a very nice result in this way, see my answer.
|
|
| Feb 20, 2017 at 12:51 | comment | added | Szabolcs | Yes, the two endpoints will always be fixed and cannot be removed. But it is always only two points, even for a letter Z. This means that the top and bottom segment of the Z may each use 3 points instead of only two endpoints. Thus there will be some redundancy, but not much. Your angle based method will not have this problem. | |
| Feb 20, 2017 at 12:46 | comment | added | Alexey Popkov | With this method the topmost and bottommost points are fixed and can't be removed. Consider the case of glyph "Z": after converting into boundary mesh it'll have many redundant topmost and bottommost points with identical $y$ coordinates... The same is true for many other glyphs. | |
| Feb 20, 2017 at 12:11 | comment | added | Szabolcs | @AlexeyPopkov Maybe something like this. It splits the closed curve into two open curves at the topmost and bottommost (ymax, ymin) points, and applies simplification separately. | |
| Feb 20, 2017 at 11:05 | comment | added | Alexey Popkov | You mean, two overlapping segments? Without overlaps we'll get fixed boundary points again. | |
| Feb 20, 2017 at 10:44 | comment | added | Szabolcs | @AlexeyPopkov Yes, that is correct. It might be better to split the data into two segments and apply the algorithm separately. | |
| Feb 20, 2017 at 8:14 | comment | added | Alexey Popkov |
As I understand, SimplifyLine doesn't take into account that we have closed curve and should treat list of points as cyclic, it never removes first and last points.
|
|
| Feb 18, 2017 at 2:49 | comment | added | Alexey Popkov | Related Demonstration by Mark McClure: demonstrations.wolfram.com/PolylineSimplification | |
| Feb 16, 2017 at 19:18 | history | edited | Szabolcs | CC BY-SA 3.0 |
added 680 characters in body
|
| Feb 16, 2017 at 19:09 | history | answered | Szabolcs | CC BY-SA 3.0 |