Any good way to determine coplanarity among 4 points?

  • Thread starter MRT
  • Start date
  • #1
MRT
5
0

Main Question or Discussion Point

Hi,

Can someone suggest a method? I need to implement it on computer.

Thanks in advance!
 

Answers and Replies

  • #2
robphy
Science Advisor
Homework Helper
Insights Author
Gold Member
5,633
890
Choose one point as an origin. Form three displacement vectors from that point. Calculate the scalar triple product (the determinant in 3d). Coplanar if zero.
 
  • #3
MRT
5
0
Hi robphy,
thanks for ur help. But the problem is that i'm working with 2-d images. So there is no 3D information.

is there any other way of determining coplanarity?
I read something about using invariants but i have great difficulty understanding the paper. The link is here below:
http://www-users.cs.york.ac.uk/~nep/research/iros01.pdf [Broken]
under section 2.2.

If someone can explain to me how to determine the point k, j, i,l in that section, i will really appreciate your help.

Thanks in advance!
 
Last edited by a moderator:
  • #4
HallsofIvy
Science Advisor
Homework Helper
41,833
956
MRT said:
Hi robphy,
thanks for ur help. But the problem is that i'm working with 2-d images. So there is no 3D information.
Perhaps it would help if you would tell us what in the world you mean by "coplanarity" in 2d?? If you have no 3D information then there is no way to determine 3D properties.
 
  • #5
MRT
5
0
Sorry for not giving a clear idea of my problem.


The problem is as follows:

I have a 2D image of a scene. I classify the scene as ground plane and non-ground plane. I have 4 points lying on these ground plane. For each point, there will be a value (x, y), with x and y representing its pixel location.

Is there a way to determine if these 4 points lie on the same plane with just this 2-D information?

Thanks in advance again!!
 
  • #6
jim mcnamara
Mentor
3,918
2,312
You can't do that. 2d representations of 3d are missing information - specifically the "z" coordinate, if you chose {x,y,z} for your system.

You're obviously writing graphics code. Rather than stating it the way you did, what are you really trying to do? ie., what is your goal?
 
  • #7
MRT
5
0
I'm actually trying to find the homography matrix H for the ground plane. I need this matrix H to classify non-ground and ground pixels. The eventual goal is to build a navigation map based on camera capture of the surrounding environment.
:smile:
 
Last edited:
  • #8
Tide
Science Advisor
Homework Helper
3,076
0
I suggest using two cameras at two locations to provide two images of the same scene.
 
  • #9
MRT
5
0
sorry. i forgot one more constraint. I'm working with only one camera. :)
 
  • #10
rachmaninoff
Then you don't have any depth perception.
 
  • #11
Tide
Science Advisor
Homework Helper
3,076
0
MRT said:
sorry. i forgot one more constraint. I'm working with only one camera. :)
Can you take two pictures with the same camera from two different points? Otherwise, unless you have objects of known size, position and orientation in your single picture, you are out of luck! :)
 

Related Threads on Any good way to determine coplanarity among 4 points?

  • Last Post
Replies
4
Views
1K
Replies
35
Views
32K
Replies
1
Views
527
Replies
1
Views
454
Replies
3
Views
18K
Replies
3
Views
2K
Replies
3
Views
1K
  • Last Post
Replies
3
Views
5K
Top