Finding concave hull around abstract shape

28 visualizaciones (últimos 30 días)
Nicolas Kaiser
Nicolas Kaiser el 10 de Oct. de 2024
Comentada: Nicolas Kaiser el 28 de Oct. de 2024 a las 7:24
Hallo together,
I have quite some abstract shape and try to found a more rectangular concave hull around it. I attached the polyshape data.
Now I wanna achieve a boundary faced like this and generate a patch object out of it:
I tried already a lot. Using the boundary function or alphaShape, but I could never reach a similar results. Either I had hole in my polyshape object at the end, or there was to much area close to the circle attached to the new polyshape object. Of course the with red indicated areas to be added to the final polyshape object dont have to be exactly like in the picture but similar I wanna achieve.
Also similar shapes should be solvelable.
Does somebody have a idear how to solve this? I would be really glad.
Thanks a lot!

Respuestas (3)

Star Strider
Star Strider el 10 de Oct. de 2024
I am not certain what you want. The information creates a 2D polyshape.
The boundary function is likely most appropriate here, although you will have to experiment with various values of the shrink factor to get the result you want.
Try this —
LD = load('matlab_answer_example.mat')
LD = struct with fields:
new_polyShape1: [1x1 polyshape]
ps = LD.new_polyShape1
ps =
polyshape with properties: Vertices: [722x2 double] NumRegions: 1 NumHoles: 0
xv = ps.Vertices(:,1);
yv = ps.Vertices(:,2);
[ixb1,v1] = boundary(xv, yv, 0.1)
ixb1 = 211×1
1 722 721 720 719 718 717 716 715 714
<mw-icon class=""></mw-icon>
<mw-icon class=""></mw-icon>
v1 = 2.5577e+03
[ixb2,v1] = boundary(xv, yv, 0.8)
ixb2 = 240×1
1 722 721 720 719 718 717 716 715 714
<mw-icon class=""></mw-icon>
<mw-icon class=""></mw-icon>
v1 = 821.0291
figure
plot(ps, 'DisplayName','Original Polyshape')
hold on
plot(xv(ixb1), yv(ixb1), '-c', 'DisplayName','Shrink = 0.1')
plot(xv(ixb2), yv(ixb2), '-m', 'DisplayName','Shrink = 0.8')
hold off
axis('equal')
legend('Location','best')
.

Matt J
Matt J el 10 de Oct. de 2024
Editada: Matt J el 11 de Oct. de 2024
ps = load('matlab_answer_example.mat').new_polyShape1;
tic;
V=cell2mat(arrayfun(@(i) upsamp(polybuffer(ps,-i)),[0,1],'UniformOutput',0)');
shp=alphaShape(V,6,'HoleThreshold',2*area(ps));
toc
Elapsed time is 1.110689 seconds.
plot(shp,'EdgeColor','none');hold on
plot(ps, 'FaceColor','none','EdgeColor','red');hold off
axis equal
function V=upsamp(ps)
V=ps.Vertices;
V(any(~isfinite(V),2),:)=[];
V(end+1,:)=V(1,:);
N=height(V);
V=interp1(V,linspace(1,N,100*N));
V(end,:)=[];
end
  2 comentarios
Nicolas Kaiser
Nicolas Kaiser el 11 de Oct. de 2024
Thanky Matt! That looks for this case already like what I wanted to achieve. I will check if I can adjust it also to different case. Maybe also to find in general a better approach.
Matt J
Matt J el 11 de Oct. de 2024
I've modified it slightly to make it faster.

Iniciar sesión para comentar.


John D'Errico
John D'Errico el 10 de Oct. de 2024
Editada: John D'Errico el 10 de Oct. de 2024
I don't know the soure of this data. But, can you start with the two pieces as separate entities, form polyshapes from each, and then use the union tool? For example, here I'll separate the two into pieces artificially.
load matlab_answer_example.mat
I1 = 70:570;
I2 = [1:70,571:722];
xy = new_polyShape1.Vertices;
S1 = polyshape(xy(I1,:));
S2 = polyshape(xy(I2,:));
Warning: Polyshape has duplicate vertices, intersections, or other inconsistencies that may produce inaccurate or unexpected results. Input data has been modified to create a well-defined polyshape.
H1 = plot(S1);
H1.FaceColor = 'r';
hold on
H2 = plot(S2);
H2.FaceColor = 'b';
hold off
S3 = union(S1,S2);
plot(S3)
  3 comentarios
Steven Lord
Steven Lord el 11 de Oct. de 2024
What specifically do you expect the two polyshapes to look like in this case after you've adjusted them?
clear all
load MatlabAnswer_polyShapes.mat
whos
Name Size Bytes Class Attributes polyShape_1 1x1 3529 polyshape polyShape_2 1x1 3497 polyshape polyShapes_1_CutOut 8x1 40136 polyshape polyShapes_2_CutOut 4x1 20068 polyshape
plot(polyShapes_1_CutOut, FaceColor='r')
hold on
plot(polyShapes_2_CutOut, FaceColor='c')
Do you want one of the polyshapes to contain only the cyan circles and the other only the red?
figure
plot(polyShapes_1_CutOut, FaceColor='r')
hold on
plot(polyShapes_2_CutOut, FaceColor='c')
plot(polyShape_1, FaceColor = 'none', EdgeColor = 'r')
plot(polyShape_2, FaceColor = 'none', EdgeColor = 'c')
polyShape_3 = subtract(polyShape_2,polyShape_1);
plot(polyShape_3, FaceColor = 'none', EdgeColor = 'k')
The subtract function got part of the way there, though the edge goes through one of the cyan circles. Is this on the right track for what you're hoping to create?
And just to confirm that you mean what I think you mean, when you say "Additionaly I want, that both polyshapes have a distance of x (x=1) towards each other." do you mean that you want the shortest distance between a point on the polyshape surrounding the cyan circles and a point on the polyshape surrounding the red circles to be 1 unit (exactly 1 unit or at least 1 unit?) So if we applied polybuffer with a distance of 1 to one of the polyshapes, it would not overlap (or not even touch) the other?
Nicolas Kaiser
Nicolas Kaiser el 28 de Oct. de 2024 a las 7:24
Sorry for the late response. You summarized it quite well. So one polyshapes should include the cyan coloured circles and the other one the red coloured circles. Exactly if we apply polybuffer to one of the polyshapes with a distance of 1 they should touch but should not overlapp (so exactly 1).
In the end I want to gain a polyshape that doesn't have the exact boundary the subtraction of the overlapping area from both polyshapes would have, to prevent a polyshape like in the post of John D'Errico but more like one in the post of Matt J. So to speak I want the resulting polyshape to be more straight and "smooth".

Iniciar sesión para comentar.

Categorías

Más información sobre Elementary Polygons en Help Center y File Exchange.

Productos


Versión

R2023b

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by