Why is the transpose of N necessary to get the point?
sushain
I think the transpose is just to make the product typecheck. That is, since points are stored as row vectors, one must be transposed into a column to enable taking a product.
azwarens
Can someone explain the +tN portion of this?
kayvonf
@sushain. Correct y^Tx is the matrix product of the 1x3 vector y^T and 3x1 vector x, which is the dot product.
kayvonf
@azwarens: Remember, points on the line are all points x such that N^Tx = c.
The direction from P directly toward the line is the opposite of the normal: -N. Therefore the value of t such that x=P-tN produces a point that satisfies N^Tx = c gives the distance from the point to the line.
tap22sf
So does this mean that N^T(p+tN)=c should be N^T(p-tN)=c in the slides above?
kayvonf
@tap22sf. Either is fine. Note that with the equation on the slide, you'd just solve for a negative value of t. However, I agree that it would have been a little more clear had I gone with the P-tN formulation on the slide.
csciutto
Wouldn't $$d = |(p - x_0)^TN|$$ be a cleaner solution? We just project the $p$ vector onto the normal and check its magnitude.
m11
What exactly is the x vector in this equation representing?
imm
@m11 I think the x vector represents points on the line represented as vectors. So any x such that NˆTx=c is on the line
azwarens
didn't get email notification...@kayvonf thank you that makes sense!
ntm
Just to check my understanding of this thread, does t represent the distance between p and the line, and does p - tN equal the "closest point" on the line?
Why is the transpose of N necessary to get the point?
I think the transpose is just to make the product typecheck. That is, since points are stored as row vectors, one must be transposed into a column to enable taking a product.
Can someone explain the +tN portion of this?
@sushain. Correct
y^Tx
is the matrix product of the 1x3 vectory^T
and 3x1 vectorx
, which is the dot product.@azwarens: Remember, points on the line are all points
x
such thatN^Tx = c
.The direction from
P
directly toward the line is the opposite of the normal:-N
. Therefore the value oft
such thatx=P-tN
produces a point that satisfiesN^Tx = c
gives the distance from the point to the line.So does this mean that N^T(p+tN)=c should be N^T(p-tN)=c in the slides above?
@tap22sf. Either is fine. Note that with the equation on the slide, you'd just solve for a negative value of t. However, I agree that it would have been a little more clear had I gone with the P-tN formulation on the slide.
Wouldn't $$d = |(p - x_0)^TN|$$ be a cleaner solution? We just project the $p$ vector onto the normal and check its magnitude.
What exactly is the x vector in this equation representing?
@m11 I think the x vector represents points on the line represented as vectors. So any x such that NˆTx=c is on the line
didn't get email notification...@kayvonf thank you that makes sense!
Just to check my understanding of this thread, does
t
represent the distance betweenp
and the line, and doesp - tN
equal the "closest point" on the line?