Article · Wikipedia archive · Last revised Jun 15, 2026

Oriented projective geometry

Oriented projective geometry is an oriented version of real projective geometry.

Last revised
Jun 15, 2026
Read time
≈ 4 min
Length
872 w
Citations
6
Source

Oriented projective geometry is an oriented version of real projective geometry.

Whereas the real projective plane describes the set of all unoriented lines through the origin in R3, the oriented projective plane describes lines with a given orientation. There are applications in computer graphics and computer vision where it is necessary to distinguish between rays light being emitted or absorbed by a point.

Elements in an oriented projective space are defined using signed homogeneous coordinates. Let R n {\displaystyle \mathbb {R} _{*}^{n}} be the set of elements of R n {\displaystyle \mathbb {R} ^{n}} excluding the origin.

  1. Oriented projective line, T 1 {\displaystyle \mathbb {T} ^{1}} : ( x , w ) R 2 {\displaystyle (x,w)\in \mathbb {R} _{*}^{2}} , with the equivalence relation ( x , w ) ( a x , a w ) {\displaystyle (x,w)\sim (ax,aw)\,} for all a > 0 {\displaystyle a>0} .
  2. Oriented projective plane, T 2 {\displaystyle \mathbb {T} ^{2}} : ( x , y , w ) R 3 {\displaystyle (x,y,w)\in \mathbb {R} _{*}^{3}} , with ( x , y , w ) ( a x , a y , a w ) {\displaystyle (x,y,w)\sim (ax,ay,aw)\,} for all a > 0 {\displaystyle a>0} .

These spaces can be viewed as extensions of euclidean space. T 1 {\displaystyle \mathbb {T} ^{1}} can be viewed as the union of two copies of R {\displaystyle \mathbb {R} } , the sets (x,1) and (x,-1), plus two additional points at infinity, (1,0) and (-1,0). Likewise T 2 {\displaystyle \mathbb {T} ^{2}} can be viewed as two copies of R 2 {\displaystyle \mathbb {R} ^{2}} , (x,y,1) and (x,y,-1), plus one copy of T {\displaystyle \mathbb {T} } (x,y,0).

An alternative way to view the spaces is as points on the circle or sphere, given by the points (x,y,w) with

x2+y2+w2=1.

Oriented real projective space

Let n be a nonnegative integer. The (analytical model of, or canonical1) oriented (real) projective space or (canonical2) two-sided projective3 space T n {\displaystyle \mathbb {T} ^{n}} is defined as

T n = { { λ Z : λ R > 0 } : Z R n + 1 { 0 } } = { R > 0 Z : Z R n + 1 { 0 } } . {\displaystyle \mathbb {T} ^{n}=\{\{\lambda Z:\lambda \in \mathbb {R} _{>0}\}:Z\in \mathbb {R} ^{n+1}\setminus \{0\}\}=\{\mathbb {R} _{>0}Z:Z\in \mathbb {R} ^{n+1}\setminus \{0\}\}.} 4

Here, we use T {\displaystyle \mathbb {T} } to stand for two-sided.

Distance in oriented real projective space

The Euclidean distance between two points p = ( p x , p y , p w ) {\displaystyle p=(p_{x},p_{y},p_{w})} and q = ( q x , q y , q w ) {\displaystyle q=(q_{x},q_{y},q_{w})} in T 2 {\displaystyle \mathbb {T} ^{2}} can be defined as elements

( ( p x q w q x p w ) 2 + ( p y q w q y p w ) 2 , p w q w ) ) {\displaystyle ({\sqrt {(p_{x}q_{w}-q_{x}p_{w})^{2}+(p_{y}q_{w}-q_{y}p_{w})^{2}}},p_{w}q_{w}))}

in T 1 {\displaystyle \mathbb {T} ^{1}} .5

The squared Euclidean distance is then:

( ( p x q w q x p w ) 2 + ( p y q w q y p w ) 2 , s i g n ( p w q w ) ( p w q w ) 2 ) {\displaystyle ((p_{x}q_{w}-q_{x}p_{w})^{2}+(p_{y}q_{w}-q_{y}p_{w})^{2},\mathrm {sign} (p_{w}q_{w})(p_{w}q_{w})^{2})}

Oriented complex projective geometry

Let n be a nonnegative integer. The oriented complex projective space C P S 1 n {\displaystyle {\mathbb {CP} }_{S^{1}}^{n}} is defined as

C P S 1 n = { { λ Z : λ R > 0 } : Z C n + 1 { 0 } } = { R > 0 Z : Z C n + 1 { 0 } } {\displaystyle {\mathbb {CP} }_{S^{1}}^{n}=\{\{\lambda Z:\lambda \in \mathbb {R} _{>0}\}:Z\in \mathbb {C} ^{n+1}\setminus \{0\}\}=\{\mathbb {R} _{>0}Z:Z\in \mathbb {C} ^{n+1}\setminus \{0\}\}} .6 Here, we write S 1 {\displaystyle S^{1}} to stand for the 1-sphere.
See also

See also

Notes

Notes

  1. Stolfi 1991, p. 2.
  2. Stolfi 1991, p. 13.
  3. Werner 2003.
  4. Yamaguchi 2002, pp. 33–34, Definition 4.1.
  5. Stolfi 1991, §17.4.
  6. Below 2003.
References

References