I have a set of points (let's say 10000 points), which are detected with much time.

Then, some points (let's say 100 points) are quickly detected, which should be included by the 10000 points. The quick method may influence by noise. Thus, there may be 2~3 points is not included by the 10000 points.

Now, I want to match the 100 points with the 10000 points. Only coordinate of points is known.

I have tried the point cloud registration method. But it cost too long time, and often failed.

Is there any other method which is faster and more robust?

Does graph method work? For example, the Ullmann method.

I have a set of points (let's say 10000 points), which are detected with much time.

Then, some points (let's say 100 points) are quickly detected, which should be included by the 10000 points. The quick method may influence by noise. Thus, there may be 2~3 points is not included by the 10000 points.

Now, I want to match the 100 points with the 10000 points. Only coordinate of points is known.

I have tried the point cloud registration method. But it cost too long time, and often failed.

Is there any other method which is faster and more robust?

Does graph method work? For example, the Ullmann method.

• math
• graph
• geometry
• registration
• point-clouds

Share Improve this question Follow edited Nov 19, 2024 at 13:48 halfer 20.4k1919 gold badges109109 silver badges202202 bronze badges asked Nov 19, 2024 at 13:47 Qiang ZhangQiang Zhang 9521313 silver badges3939 bronze badges 4

• "I have tried the point cloud registration method" - Which one? There many methods for this problem. "For example the Ullmann method" - Are you referring to a specific software library here? – IBitMyBytes Commented Nov 19, 2024 at 14:21
• Thanks for your kindly reply. I have tried the point cloud registration method: ICP. It is not robust. The Ullmann method is not a software library. It is a graph method, which can match a sub-graph with a graph. But the Ullmann method requires that the point of sub-graph should must included by the whole graph. So, I doubt whether the Ullmann works for point cloud registration. In the other hand, do you know some other faster and more rubost point cloud registration method? – Qiang Zhang Commented Nov 20, 2024 at 0:21
• Are you looking for methods/algorithms in general or for an implemented method/software library? If it is the second one, which programming language? – IBitMyBytes Commented Nov 20, 2024 at 8:52
• I am looking for algorithms in general. Not to implement software library. I want to use c++ to implement it in my project. – Qiang Zhang Commented Nov 20, 2024 at 14:02

Add a comment  | 

1 Answer 1

Sorted by: Reset to default 0

Here are two methods you could look into:

• https://dl.acm./doi/10.1145/1360612.1360684
• https://isprs-annals.copernicus./articles/II-5-W2/283/2013/

The second one builds upon the first one. They have the advantage that they do not need an initial guess of the transformation between the two point clouds (in contrast to e.g. ICP).