VTK 检查多数据点对象的交叉点

Question

我有两个 VTK 多数据对象，其中包含要检查交叉点的点数据。本质上，我想将这些点做成多边形，然后使用它。

我被告知这样做的方法是convert the point data into a triangular mesh，然后使用vtkIntersectionPolyDataFilter 进行检查。这是我目前拥有的：

def convert_pts_to_mesh(polydata):
    aCellArray = vtk.vtkCellArray()

    boundary = vtk.vtkPolyData()
    boundary.SetPoints(polydata.GetPoints())
    boundary.SetPolys(aCellArray)
    delaunay = vtk.vtkDelaunay2D()
    delaunay.SetInputData(polydata)
    delaunay.SetSourceData(boundary)
    delaunay.Update()

    result_polydata = delaunay.GetOutput()

    # print("result_polydata:")
    # print(result_polydata)
    return result_polydata

...

contour1 = ... # Source of polydata point object
contour2 = ... # Source of polydata point object

# Convert them to triangle meshes.
result_polydata1 = convert_pts_to_mesh(contour1)
result_polydata2 = convert_pts_to_mesh(contour2)

intersection_operation = vtk.vtkIntersectionPolyDataFilter()
intersection_operation.SetInputData(0, result_polydata1)
intersection_operation.SetInputData(1, result_polydata2)
intersection_operation.Update()

print("# of crosses: " + str(intersection_operation.GetNumberOfIntersectionPoints()))

但是，这会吐出我认为与 intersection_operation.Update() 调用中的失败有关的错误。

#121.040# [VtkError] ERROR: ERROR: In /usr/local/sv/ext/2019.02/release/gl2/src/vtk-8.1.1/Common/DataModel/vtkPointLocator.cxx, line 867
vtkPointLocator (0x555d26925800): No points to subdivide
!121.041! [VtkGenericWarning] WARNING: Generic Warning: In /usr/local/sv/ext/2019.02/release/gl2/src/vtk-8.1.1/Filters/General/vtkIntersectionPolyDataFilter.cxx, line 2518
No Intersection between objects
# of crosses: 0

事实上，错误提到了细分点，我尝试输入 contour1 和 contour2 对象，但随后在 SetInputData 行上出错：

!126.179! [VtkGenericWarning] WARNING: Generic Warning: In /usr/local/sv/ext/2019.02/release/gl2/src/vtk-8.1.1/Common/Core/vtkMath.cxx, line 779
vtkMath::Jacobi: Error extracting eigenfunctions

我不知道从哪里开始，Delaunay 和 IntersectonPolyDataFilter 上的 VTK 文档对我来说并不是最有用的。

Answer 1

如果你所说的交点是指点云的重叠，你可以试试：

import numpy as np
from vedo import Points, ConvexHull, show

pts = np.random.rand(1000, 3)

# Points() creates a vtkActor (with extended functionalities) from
# the original cloud of points pts:
pts1 = Points(pts, r=5).c('red') # r=radius of points

# create a second cloud displaced by a constant
pts2 = Points(pts+[0.6,0.8,0], r=2).c('green')

# create the convex hull which wraps the pts2 cloud.
# ch2 is also a vtkActor. Note that you can concatenate commands
# like .alpha() and .c() to change transparency and color..
ch2 = ConvexHull(pts2).alpha(0.2).c('green')

# extract the points of the original cloud (pts) which are inside
# the convex hull of the second (ch2) as a list of points:
print("# of crosses:", len(ch2.insidePoints(pts).points()))

# show() will render all the indicated objects:
show(pts1, pts2, ch2, axes=1)