home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.graphics.visualization
- Path: sparky!uunet!caen!med.umich.edu!spencer
- From: spencer@med.umich.edu (Spencer W. Thomas)
- Subject: Re: Curvature of a Surface
- In-Reply-To: srinivas@lgc.com's message of Thu, 27 Aug 1992 05:45:17 GMT
- Message-ID: <SPENCER.92Aug27135113@guraldi.med.umich.edu>
- Followup-To: comp.graphics.visualization
- Sender: news@med.umich.edu
- Organization: University of Michigan
- References: <1992Aug27.054517.4151@lgc.com>
- Date: Thu, 27 Aug 1992 18:51:13 GMT
- Lines: 51
-
- There is no single number that is the "curvature" of a curface.
- Directional curvature is defined as the curvature along a particular
- tangent direction, and is the curvature of the curve of intersection
- between a plane perpendicular to the surface at the point. The
- minimum and maximum values of the directional curvature occur along
- perpendicular directions and are called the principal curvatures. The
- Gaussian curvature is the product of the principal curvatures. It is
- positive for "sphere-like" points, negative at "saddle" points, and
- zero for ruled surfaces. The mean curvature is the average of the
- principal curvatures.
-
- To compute the curvature tensor (from which the others may be
- derived),
-
- Note: below, i,j,k are in the set {1,2} and denote the two parameters of
- the surface u1 and u2. I.e., S(u1,u2) = (x(u1,u2), y(u1,u2), z(u1,u2)).
-
- Capital letters are vector valued, lowercase are scalar (except g, g',
- L and C, which are matrices, while gij, etc. are elements of those
- matrices.)
-
- 1. Compute basic tangent vectors Xi = dS/dui
-
- 2. Compute metric coefficients gij = Xi . Xj (dot product)
-
- 3. Compute inverse metric coefficients g'ij = matrix inverse of gij
-
- 4. Compute normal vector N = Xi * Xj (cross product)
-
- 5. Compute derivatives of basic tangent vectors Xij = (d^2)S/(duj dui)
- i.e., Xij = dXi/duj
-
- 6. Compute coefficients of the second fundamental form
- Lij = Xij . N
-
- 7. Compute components of curvature tensor Cij = sum(k)(Lki g'kj)
-
- The Gaussian curvature is the determinant of the matrix C. The mean
- curvature is 1/2 trace(C). The principal curvatures (c1, c2) are the
- eigenvalues of C, and the principal directions are the corresponding
- eigenvectors. If X(1) is the first principal direction, corresponding
- to the curvature c1, then the directional curvature in the direction Y
- is c1 cos^2(theta) + c2 sin^2(theta), where theta is the angle between
- Y and X(1).
-
- Reference: Elements of Differential Geometry, Millman and Parker,
- Prentice-Hall, 1977.
-
- --
- =Spencer W. Thomas HSITN, U of Michigan, Ann Arbor, MI 48109
- spencer.thomas@med.umich.edu 313-747-2778
-