I'm writing a data analysis program and part of it requires finding the volume of a shape. The shape information comes in the form of a lost of points, giving the radius and the angular coordinates of the point.
If the data points were uniformly distributed in coordinate space I would be able to perform the integral, but unfortunately the data points are basically randomly distributed.
My inefficient approach would be to find the nearest neighbours to each point and stitch the shape together like that, finding the volume of the stitched together parts.
Does anyone have a better approach to take?
Thanks.
IF those are surface points, one good way to do it would be to discretize the surface as triangles and convert the volume integral to a surface integral using Green's Theorem. Then you can use simple Gauss quadrature over the triangles.
Ok, here it is, along duffymo's lines I think.
First, triangulate the surface, and make sure you have consistent orientation of the triangles. Meaning that orientation of neighbouring triangle is such that the common edge is traversed in opposite directions.
Second, for each triangle ABC compute this expression: H*cross2D(B-A,C-A), where cross2D computes cross product using coordinates X and Y only, ignoring the Z coordinates, and H is the Z-coordinate of any convenient point in the triangle (although the barycentre would improve precision).
Third, sum up all the above expressions. The result would be the signed volume inside the surface (plus or minus depending on the choice of orientation).
Sounds like you want the convex hull of a point cloud. Fortunately, there are efficient ways of getting you there. Check out scipy.spatial.ConvexHull.
Related
Okay first I wasn't sure if this was better suited to the MathSO so apologies if it needs migrating.
I have a 3D grid of points (representing the centers of voxels) with pitch varying in each dimension, but regular. For example resolution may be 100 by 50 by 40 for a cube shaped object.
Giving me nVox = 200,000.
For each voxel - I would like to cast (nVox - 1) rays, ending at the center, and originating from each of the other voxels.
Now there is obviously a lot of overlap here but I am having trouble finding how to calculate the minimum set of rays required. This sounds like a problem that has an elegant solution, I am however struggling to find it.
As a start, it is obvious that you only need to compute
[nVox * (nVox - 1)] / 2
of the rays, as the other half will simply be in the opposite directions. It is also easy in the 2D case to combine all of those parallel to one of the grid axes (and the two diagonals).
So how do I find the minimum set of rays I need, to pass from all voxel centers, to all others?
If someone could point me in the right direction that'd be great. Any and all help will be much appreciated.
Your problem really isn't about three dimensions in any specific way. All the conceptual complexity is present in the two dimensional case.
Instead of connecting points individually, think about the set of lines that pass through at least two points on your grid. Thus instead of thinking about points initially, think about directions. For 2-D these directions are slopes of lines. These slopes have to be rational numbers, since they intersect points on an integer lattice. Since you have a finite lattice, the numerator and the denominator of the slope can be bounded by the size of the figure. So your underlying problem is enumerating possible slopes for rational numbers of bounded "height" (math jargon).
There's an algorithm for that. It's the one used to generate the Farey sequence of reduced fractions. If your figure is N pixels wide, there will (in general) be a slope with denominator N in the somewhere, but there can't be a slope in reduced form with denominator >N; it wouldn't fit.
It's easier to deal with slopes between 0 and 1 directly. You get the other directions by two operations: negating the slope and by interchanging axes. For three dimensions, you need two slopes to define a direction.
Given an arbitrary direction (no necessarily a rational one as above), there's a perpendicular linear space of dimension k-1; for 3-D that's a plane. Projecting a 3-D parallelpiped onto this plane yields a hexagon in general; two vertices project onto the interior, six project to the vertices of the hexagon.
For a given discrete direction, there's a minimal bounding box on the integer lattice such that two opposite vertices lie along that direction. As long as that bounding box fits within your original grid, each of the interior points of the projection each correspond to a line that intersects your grid in at least two points.
In summary, enumerate directions, then for each direction enumerate where that direction intersects your grid in at least two points.
I am working on a project where I have a set of known measurements (x,y,z,a) and an input (z,a). I need to be able to interpolate the (x,y,z) so that I can get a list of possible (x,y) coordinates from a given z.
I was looking at bicubic interpolation, but I can only find examples pertaining to regular grids, and my (x,y) pairs are most certainly not regular.
Basically I am looking for some guidance on algorithms/models to achieve this goal. I am considering a triangulated irregular network, which is attractive because it breaks down into planes which are easy to determine the (x,y) from a given Z. But I would like a little more finesse.
I know it sounds like homework, its not.
Efficiency is not a concern.
Thanks!
I actually ended up using Delauney Triangulation to break down the fields into 3 dimensional X,Y,Z surfaces with an Identifier. Then given a set of (Identity,Z) pairs I form a field line from each surface, and from these lines compute the polygon formed from the shortest edges between lines. This gives me an area of potential x,y coordinates.
Take a look at Kd-tree.
These first take a set of scattered points in 2d or 3d or 10d,
then answers queries like "find the 3 points nearest P".
Are your queries z a pairs ?
For example, given a bunch of colored pins on a map, a table of x y size color,
one could put all the [x y] in a kd tree, then ask for pins near a given x0 y0.
Or, one could put all the [size color[ in a tree, then ask for pins with a similar size and color.
(Note that most kd-tree implementations use the Euclidean metric,
so sqrt( (size - size2)^2 + (color - color2)^2 ) should make sense.)
In Python, I highly recommend scipy.spatial.cKDTree.
See also SO questions/tagged/kdtree .
I am trying to find sphere that surly encompasses given list of points.
Points will have x, y and z co-ordinate[Points are in 3D].
Actually I am trying to find new three points based on given list of points by some calculations like find MinX,MaxX ,MinY,MaxY,and MinZ and MaxZ and do some operation and find new three points
And I will draw sphere from these three points.
And I will also taking all these three points on the diameter of sphere so I have a unique sphere.
Is there any standard way for finding encompassing sphere of given list of points?
Yes, the standard algorithm is Welzl's algorithm (assuming you want the minimal sphere around your points). Particularly the improved version of Gaertner is very useful, robust and numerically stable! It handles all the degenerate cases well too.
At its core, the algorithm permutes the points (randomly) to find the 1-4 points that lie on the boundary of the sphere. It's basically a clever trial-and-error algorithm. From these points, you can find the center by finding a point that has the same distance to all those points. Gärtner's version uses an improved numerical device to find the center. Also, it employs an extra pivoting step that presumably makes the algorithm work better for a large number of input points.
If all you want is a sphere around three points, I suggest you still use Gärtners "device" to compute the circumsphere of the triangle. Otherwise, the method will probably degenerate easily (i.e. when the triangle is very flat).
Do you need 3 points, or any number of points?
If you only need the answer for 3 points, each pair of points defines a line segment. Take the longest line segment. Take a sphere centered at the middle of that line segment, whose radius is half the length of the line segment. There are two cases.
The third point is inside of that initial sphere. If so, then you have the smallest sphere.
The third point is outside of that initial sphere. Then the solution at Find Circum Center of Three point of Triangle [Not using Compass] will give you the center of the smallest sphere containing those 3 points.
If you need an arbitrary number of points, I'd do some sort of iterative approximation algorithm. Since you don't seem like you need that, I won't work out the details.
I need to convert arbitrary triangulated 3D mesh to cloud of particles that are uniformly spaced.
First thought was to try find a way to fill one 3D triangle. And then fill each triangle of mesh, removing duplicated particles on edges, but that's just hard and too much work. I was hoping for some more-math way.
Can anyone point me to an algorithm which can help me do my task correctly... well, at least approximatively?
Thanks
There are two main options:
Voxelization of mesh. Easy to implement the conversion of mesh to voxels, but it's inaccurate since uniform spacing cannot be achieved: distance between cubes can be x, x*sqrt(2) or x*sqrt(3) depending if neighbor cubes are in same plane and adjacent.
Poisson disk sampling on surface. Hard to implement and lack of research material and code, but mathematically very correct. Some links:
http://research.microsoft.com/apps/pubs/default.aspx?id=135760
http://web.mysites.ntu.edu.sg/cwfu/public/Shared%20Documents/dualtiling/index.html
You could convert the TIN to raster using a GIS package or software such as R, then retrieve one point at the center of each pixel representing the value. (Example in ArcGIS)
EDIT: If the irregular 3D mesh has multiple heights per {x, y} a similar approach would be to sample the mesh using a voxel "grid" and keep one value per voxel. GRASS GIS has the functionality to take the vertices of the TIN (3d mesh) and convert them to voxels, then back to a regular 3d cloud.
I'm looking for a way to determine the optimal X/Y/Z rotation of a set of vertices for rendering (using the X/Y coordinates, ignoring Z) on a 2D canvas.
I've had a couple of ideas, one being pure brute-force involving performing a 3-dimensional loop ranging from 0..359 (either in steps of 1 or more, depending on results/speed requirements) on the set of vertices, measuring the difference between the min/max on both X/Y axis, storing the highest results/rotation pairs and using the most effective pair.
The second idea would be to determine the two points with the greatest distance between them in Euclidean distance, calculate the angle required to rotate the 'path' between these two points to lay along the X axis (again, we're ignoring the Z axis, so the depth within the result would not matter) and then repeating several times. The problem I can see with this is first by repeating it we may be overriding our previous rotation with a new rotation, and that the original/subsequent rotation may not neccesarily result in the greatest 2D area used. The second issue being if we use a single iteration, then the same problem occurs - the two points furthest apart may not have other poitns aligned along the same 'path', and as such we will probably not get an optimal rotation for a 2D project.
Using the second idea, perhaps using the first say 3 iterations, storing the required rotation angle, and averaging across the 3 would return a more accurate result, as it is taking into account not just a single rotation but the top 3 'pairs'.
Please, rip these ideas apart, give insight of your own. I'm intreaged to see what solutions you all may have, or algorithms unknown to me you may quote.
I would compute the principal axes of inertia, and take the axis vector v with highest corresponding moment. I would then rotate the vertices to align v with the z-axis. Let me know if you want more details about how to go about this.
Intuitively, this finds the axis about which it's hardest to rotate the points, ie, around which the vertices are the most "spread out".
Without a concrete definition of what you consider optimal, it's impossible to say how well this method performs. However, it has a few desirable properties:
If the vertices are coplanar, this method is optimal in that it will always align that plane with the x-y plane.
If the vertices are arranged into a rectangular box, the box's shortest dimension gets aligned to the z-axis.
EDIT: Here's more detailed information about how to implement this approach.
First, assign a mass to each vertex. I'll discuss options for how to do this below.
Next, compute the center of mass of your set of vertices. Then translate all of your vertices by -1 times the center of mass, so that the new center of mass is now (0,0,0).
Compute the moment of inertia tensor. This is a 3x3 matrix whose entries are given by formulas you can find on Wikipedia. The formulas depend only on the vertex positions and the masses you assigned them.
Now you need to diagonalize the inertia tensor. Since it is symmetric positive-definite, it is possible to do this by finding its eigenvectors and eigenvalues. Unfortunately, numerical algorithms for finding these tend to be complicated; the most direct approach requires finding the roots of a cubic polynomial. However finding the eigenvalues and eigenvectors of a matrix is an extremely common problem and any linear algebra package worth its salt will come with code that can do this for you (for example, the open-source linear algebra package Eigen has SelfAdjointEigenSolver.) You might also be able to find lighter-weight code specialized to the 3x3 case on the Internet.
You now have three eigenvectors and their corresponding eigenvalues. These eigenvalues will be positive. Take the eigenvector corresponding to the largest eigenvalue; this vector points in the direction of your new z-axis.
Now, about the choice of mass. The simplest thing to do is to give all vertices a mass of 1. If all you have is a cloud of points, this is probably a good solution.
You could also set each star's mass to be its real-world mass, if you have access to that data. If you do this, the z-axis you compute will also be the axis about which the star system is (most likely) rotating.
This answer is intended to be valid only for convex polyhedra.
In http://203.208.166.84/masudhasan/cgta_silhouette.pdf you can find
"In this paper, we study how to select view points of convex polyhedra such that the silhouette satisfies certain properties. Specifically, we give algorithms to find all projections of a convex polyhedron such that a given set of edges, faces and/or vertices appear on the silhouette."
The paper is an in-depth analysis of the properties and algorithms of polyhedra projections. But it is not easy to follow, I should admit.
With that algorithm at hand, your problem is combinatorics: select all sets of possible vertexes, check whether or not exist a projection for each set, and if it does exists, calculate the area of the convex hull of the silhouette.
You did not provide the approx number of vertex. But as always, a combinatorial solution is not recommended for unbounded (aka big) quantities.