https://stackoverflow.com/questions/17775832
italiano
english
français
española
中国
日本の
العربية
Deutsch
한국어
Português
RussianUsing a 2D vector space approximation (local tangent space)
In this section, I can detail how I come to these formulas.
Let's note Points the points of the polygon (where Points[0] == Points[Points.Count - 1] to close the polygon).
The idea behind the next methods is to split the polygon into triangles (the area is the sum of all triangle areas). But, to support all polygon types with a simple decomposition (not only star-shaped polygon), some triangle contributions are negative (we have a "negative" area). The triangles decomposition I use is : {(O, Points[i], Points[i + 1]} where O is the origin of the affine space.
The area of a non-self-intersecting polygon (in euclidian geometry) is given by:
In 2D:
float GetArea(List<Vector2> points)
{
float area2 = 0;
for (int numPoint = 0; numPoint < points.Count - 1; numPoint++)
{
MyPoint point = points[numPoint];
MyPoint nextPoint = points[numPoint + 1];
area2 += point.x * nextPoint.y - point.y * nextPoint.x;
}
return area2 / 2f;
}
In 3D, given normal, the unitary normal of the polygon (which is planar):
float GetArea(List<Vector3> points, Vector3 normal)
{
Vector3 vector = Vector3.Zero;
for (int numPoint = 0; numPoint < points.Count - 1; numPoint++)
{
MyPoint point = points[numPoint];
MyPoint nextPoint = points[numPoint + 1];
vector += Vector3.CrossProduct(point, nextPoint);
}
return (1f / 2f) * Math.Abs(Vector3.DotProduct(vector, normal));
}
In the previous code I assumed you have a Vector3 struct with Add, Subtract, Multiply, CrossProduct and DotProduct operations.
In your case, you have a lattitude and longitude. Then, you are not in an 2D euclidean space. It is a spheric space where computing the area of any polygon is much more complex. However, it is locally homeomorphic to a 2D vector space (using the tangent space). Then, if the area you try to measure is not too wide (few kilometers), the above formula should work.
Now, you just have to find the normal of the polygon. To do so, and to reduce the error (because we are approximating the area), we use the normal at the centroid of the polygon. The centroid is given by:
Vector3 GetCentroid(List<Vector3> points)
{
Vector3 vector = Vector3.Zero;
Vector3 normal = Vector3.CrossProduct(points[0], points[1]); // Gets the normal of the first triangle (it is used to know if the contribution of the triangle is positive or negative)
normal = (1f / normal.Length) * normal; // Makes the vector unitary
float sumProjectedAreas = 0;
for (int numPoint = 0; numPoint < points.Count - 1; numPoint++)
{
MyPoint point = points[numPoint];
MyPoint nextPoint = points[numPoint + 1];
float triangleProjectedArea = Vector3.DotProduct(Vector3.CrossProduct(point, nextPoint), normal);
sumProjectedAreas += triangleProjectedArea;
vector += triangleProjectedArea * (point + nextPoint);
}
return (1f / (6f * sumProjectedAreas)) * vector;
}
I've added a new property to Vector3 : Vector3.Length
Finally, to convert latitude and longitude into a Vector3:
Vector3 GeographicCoordinatesToPoint(float latitude, float longitude)
{
return EarthRadius * new Vector3(Math.Cos(latitude) * Math.Cos(longitude), Math.Cos(latitude) * Math.Sin(longitude), Math.Sin(latitude));
}
To sum up:
// Converts the latitude/longitude coordinates to 3D coordinates
List<Vector3> pointsIn3D = (from point in points
select GeographicCoordinatesToPoint(point.Latitude, point.Longitude))
.ToList();
// Gets the centroid (to have the normal of the vector space)
Vector3 centroid = GetCentroid(pointsIn3D );
// As we are on a sphere, the normal at a given point is the colinear to the vector going from the center of the sphere to the point.
Vector3 normal = (1f / centroid.Length) * centroid; // We want a unitary normal.
// Finally the area is computed using:
float area = GetArea(pointsIn3D, normal);
The Vector3 struct
public struct Vector3
{
public static readonly Vector3 Zero = new Vector3(0, 0, 0);
public readonly float X;
public readonly float Y;
public readonly float Z;
public float Length { return Math.Sqrt(X * X + Y * Y + Z * Z); }
public Vector3(float x, float y, float z)
{
X = x;
Y = y;
Z = z;
}
public static Vector3 operator +(Vector3 vector1, Vector3 vector2)
{
return new Vector3(vector1.X + vector2.X, vector1.Y + vector2.Y, vector1.Z + vector2.Z);
}
public static Vector3 operator -(Vector3 vector1, Vector3 vector2)
{
return new Vector3(vector1.X - vector2.X, vector1.Y - vector2.Y, vector1.Z - vector2.Z);
}
public static Vector3 operator *(float scalar, Vector3 vector)
{
return new Vector3(scalar * vector.X, scalar * vector.Y, scalar * vector.Z);
}
public static float DotProduct(Vector3 vector1, Vector3 vector2)
{
return vector1.X * vector2.X + vector1.Y * vector2.Y + vector1.Z * vector2.Z;
}
public static Vector3 CrossProduct(Vector3 vector1, Vector3 vector2)
{
return return new Vector3(vector1.Y * vector2.Z - vector1.Z * vector2.Y,
vector1.Z * vector2.X - vector1.X * vector2.Z,
vector1.X * vector2.Y - vector1.Y * vector2.X);
}
}
