Geometry of Perspective Imaging Images of the 3-D World

Geometry of Perspective Imaging Images of the 3-D World

Geometry of perspective imaging ■ Coordinate transformations ■ Image formation ■ Vanishing points ■ Stereo imaging Image formation Images of the 3-D world ■ What is the geometry of the image of a three dimensional object? – Given a point in space, where will we see it in an image? – Given a line segment in space, what does its image look like? – Why do the images of lines that are parallel in space appear to converge to a single point in an image? ■ How can we recover information about the 3-D world from a 2-D image? – Given a point in an image, what can we say about the location of the 3-D point in space? – Are there advantages to having more than one image in recovering 3-D information? – If we know the geometry of a 3-D object, can we locate it in space (say for a robot to pick it up) from a 2-D image? Image formation Euclidean versus projective geometry ■ Euclidean geometry describes shapes “as they are” – properties of objects that are unchanged by rigid motions » lengths » angles » parallelism ■ Projective geometry describes objects “as they appear” – lengths, angles, parallelism become “distorted” when we look at objects – mathematical model for how images of the 3D world are formed Image formation Example 1 ■ Consider a set of railroad tracks – Their actual shape: » tracks are parallel » ties are perpendicular to the tracks » ties are evenly spaced along the tracks – Their appearance » tracks converge to a point on the horizon » tracks don’t meet ties at right angles » ties become closer and closer towards the horizon Image formation Example 2 ■ Corner of a room – Actual shape » three walls meeting at right angles. Total of 270o of angle. – Appearance » a point on which three lines segments are concurrent. Total angle is 360o Image formation Example 3 q ■ B appears between A and C from point p A ■ But from point 1, A appears between B C and C ■ Apparent displacement of objects due to B change in viewing position is called parallax shift p – Greeks knew that is the earth revolved around the sun there would be parallax shift of the stars C – Tycho Brahe looked for this shift, but A his instruments were not accurate B enough. Image formation Vanishing points and lines ■ To begin with, let’s assume we’re looking at a scene on the plane, σ, which is perpendicular to the image plane ρ. – To each point P on σ we associate the v C point p on ρ corresponding to the V intersection of the line CP with the plane ρ. – C is called the center of projection p ■ The point of intersection of the line P through the optical center, C, that is perpendicular to the image plane (and so Object plane, σ parallel to the object plane) is called the principal vanishing point, V. ■ The line, v, which is the intersection of the picture plane and the plane through P Image plane, ρ parallel to the object plane is called the vanishing line, or the horizon line. Image formation Vanishing points and lines ■ Our mapping between σ and ρ does not define a 1-1 correspondence. – Points on v are not the image of any point on σ – Points on m (intersection of plane ρ through C parallel to ) have no v C image on σ V ■ We’ll fix this later. ■ Add railroad tracks to σ. p P m Object plane, σ Image plane, ρ Image formation Vanishing points and lines ■ The images of all the points on L1 lie on the π plane 1 defined by L1 and C. – this is because the lines of sight for each point on v L1 lie on this plane. C ■ π V The image of L1 is then the intersection of 1 with the picture plane - l1. ■ L1 is parallel to CV since they are both perpendicular to the picture plane. l2 l1 π L2 – But, then CV must also lie on 1, since parallel lines are co-planar, and V lies on π 1. L1 ■ So, the principal vanishing point must lie on l1, Object plane, σ π since it lies on both 1 and the picture plane. ■ Similarly, V must lie on l2, the image of L2. ■ In fact, ALL lines in the object plane that are ρ perpendicular to the image plane image to lines Image plane, passing through V. – This is the image of the “point at infinity” for that set of parallel lines Image formation Vanishing points and lines ■ M1 and M2 are parallel, but not perpendicular to the image plane. ■ The image of M is the line m where the plane, i i v π , containing C and M intersects the image Λ C i i V plane. M V ■ There is a unique line, Λ, through C which is π parallel to Mi and so is this line is also on i. m2 ■ Since this line is horizontal, it also lies in the M m1 plane determined by C and v, the vanishing line. 2 M ■ Since v and Λ lie in the same plane, and are not 1 Object plane, σ parallel, they must intersect at some point Vm. Image plane, ρ Image formation Vanishing points and lines ■ π This places Vm on both i and the image plane, so it must lie on their intersection, mi. v C – So, any family of parallel lines on the Λ V V object plane will image to a set of lines M passing through a point (vanishing point) on the vanishing line. ■ As we rotate the set of lines, the vanishing m1 m2 point moves along the vanishing line. M2 ■ So even though parallel lines don’t “meet” we M1 can see where they meet in images! Object plane, σ Image plane, ρ Image formation General case ■ Generally, the vanishing line for any object plane is the line of intersection of the –image plane – plane parallel to the object plane through C (called the horizon plane). ■ The vanishing point for a line on the object plane is the intersection of the – vanishing line – plane containing the original line and the center of projection. V C v π ρ Image formation Vanishing points and lines Image formation Homogeneous coordinates ■ Classical Euclidean geometry: through any point not on a given line, there exists a unique line which is parallel to the given line. – For 2,000 years, mathematician tried to “prove” this from Euclid’s postulates. – In the early 20’th century, geometry was revolutionized when mathematicians asked: What if this were false? – That is, what if we assumed that EVERY pair of lines intersected? – To do this, we’ll have to add points and lines to the standard Euclidean plane. ■ If (x,y) are the rectangular coordinates of a point, P, and if (x1, x2, x3) are any three real numbers such that: –x1/x3 = x –x2/x3 = y then (x1, x2, x3) are a set of homogeneous coordinates for (x,y). ■ So, in particular, (x,y,1) are a set of homogeneous coordinates for (x,y) Image formation Homogeneous coordinates ■ Given the homogeneous coordinates, (x1, x2, x3), the rectangular coordinates can be recovered. ■ But (x,y) has an infinite number of homogeneous coordinate representations, because if (x1, x2, x3) are homogeneous coordinates of (x,y), then so are (kx1, kx2,k x3) for any k <>0. Image formation Points and lines ■ In rectangular coordinates, we can represent a line by: – l1x + l2y + l3 = 0 l1, l2 not both 0. ■ In homogeneous coordinates, the equation becomes – l1x1 + l2x2 + l3x3 = 0 ■ So a line is determined by the three coefficients (l1, l2, l3) ■ If l is the line l1x1 + l2x2 + l3x3 = 0, then we say that [l1, l2, l3] are the homogeneous coordinates of l. – So, in homogeneous coordinates, points and lines look the same - triples of numbers. – We’ll represent points as (…) and lines as […] to keep things straight. ■ A line is uniquely determined by a set of homogeneous coordinates, but a given line has an infinite number of homogeneous coordinate representations. –If [l1, l2, l3] represents a given line, then so does [kl1, kl2,k l3]. » I.e., the same set of points satisfy the equations l1x1 + l2x2 + l3x3 = 0 and kl1x1 +k l2x2 + kl3x3 = 0 Image formation Vectors: Inner Product ■ T Inner product of two vectors a=(ax ay az) and b=(b b b )T is a scalar b x y z α α a·b =|a|·|b|cos = axbx + ayby+ az bz |b|cosα a Note that we are projecting one vector onto the other. ■ Length: ≡ T 1/2 2 2 2 1/2 |a| ||a|| =(a ·a) = (ax + ay + az ) Image formation Vectors: Cross product ■ Vector product of two vectors a and b is a vector a×b i j k b × a b =det ax ay az = α bx by bz a = i(aybz –azby)+ j(azbx –axbz)+ k(axby –aybx) ■ This vector is orthogonal to the plane formed by a and b (in the sense of a right-handed screw) ■ Its magnitude is the area of the parallelogram whose sides are a and b |a×b| = |a|·|b| |sinα| ■ For two vectors in the xy-plane the vector product can be used to determine convexity! How? Image formation Vectors: Mixed vector product ■ Mixed vector product of two vectors a c b and b is a scalar α ax ay az × a (a b)·c =det bx by bz cx cy cz ■ Its magnitude is the volume of the parallelepiped whose sides are a, b, and c Image formation Points and lines ■ We usually think of the equation of a line as specifying the set of points that lie on that line.

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    40 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us