![]() | This article is rated Start-class on Wikipedia's
content assessment scale. It is of interest to the following WikiProjects: | ||||||||||
|
The original Horn & Shunck method was described in 2 dimensions only (x and y). I think its easier for the reader to imagine the whole thing with just two dimensions (plus time). -- Oxygene123 16:08, 31 January 2006 (UTC)
I believe there is an error in the equations, the Laplace operator should be and analogously for y and z. Also the so claimed Gauss-Seidel formulae seem wrong, as there will be division by zero wherever the derivative is zero.
In the iterative equations given, (certainly for the Gauss-Seidel method) shouldn't the newly computed values be used immediately? Pog ( talk) 10:00, 21 November 2007 (UTC)
In addition to the points mentioned above: - Nomenclature is really messy. denotes a partial differentiation while is supposed to be the first component of V. I'd prefer sticking to the variables used by Horn & Schunck. - Gauss-Seidel could be applied, but after replacing the Laplacian. This way there would be no division by zero. - The second set of equations is obtained by simply solving for u and v. I don't see the point in mentioning the Jacobi method, even though this could qualify as a trivial case of the Jacobi method. ( 85.216.75.129 ( talk) 17:29, 23 June 2008 (UTC)).
This may be a better URL for the paper http://dspace.mit.edu/handle/1721.1/6337, as it is from the same institution as the authors (MIT), and includes useful metadata and a copyright statement. The article uses a URL to a personal webpage with a scan of doubtful copyright status. — DIV ( 138.194.12.32 ( talk) 08:03, 10 August 2010 (UTC))
![]() | This article is rated Start-class on Wikipedia's
content assessment scale. It is of interest to the following WikiProjects: | ||||||||||
|
The original Horn & Shunck method was described in 2 dimensions only (x and y). I think its easier for the reader to imagine the whole thing with just two dimensions (plus time). -- Oxygene123 16:08, 31 January 2006 (UTC)
I believe there is an error in the equations, the Laplace operator should be and analogously for y and z. Also the so claimed Gauss-Seidel formulae seem wrong, as there will be division by zero wherever the derivative is zero.
In the iterative equations given, (certainly for the Gauss-Seidel method) shouldn't the newly computed values be used immediately? Pog ( talk) 10:00, 21 November 2007 (UTC)
In addition to the points mentioned above: - Nomenclature is really messy. denotes a partial differentiation while is supposed to be the first component of V. I'd prefer sticking to the variables used by Horn & Schunck. - Gauss-Seidel could be applied, but after replacing the Laplacian. This way there would be no division by zero. - The second set of equations is obtained by simply solving for u and v. I don't see the point in mentioning the Jacobi method, even though this could qualify as a trivial case of the Jacobi method. ( 85.216.75.129 ( talk) 17:29, 23 June 2008 (UTC)).
This may be a better URL for the paper http://dspace.mit.edu/handle/1721.1/6337, as it is from the same institution as the authors (MIT), and includes useful metadata and a copyright statement. The article uses a URL to a personal webpage with a scan of doubtful copyright status. — DIV ( 138.194.12.32 ( talk) 08:03, 10 August 2010 (UTC))