Lots of video-processing algorithms are 3D applications (2 spatial dimensions and one temporal dimension) of basic 1D filters used in DSP applications. Personally, I find this approach very attractive, mainly because of the elegant computational architectures that result and the potential for massive parallelism. See some of the early papers  by

https://www.researchgate.net/profile/LT_Bruton

or more recently (with 4-D extensions)

Madanayake, A.; Wijenayake, C.; Dansereau, D.G.; Gunaratne, T.K.; Bruton, L.T.; Williams, S.B., "Multidimensional (MD) Circuits and Systems for Emerging Applications Including Cognitive Radio, Radio Astronomy, Robot Vision and Imaging," Circuits and Systems Magazine, IEEE , vol.13, no.1, pp.10,43, Firstquarter 2013

I have written a few papers on this sort of thing recently that you could look at. But I start with a different type of 1D filter. 

 Dear Prof. Hugh 

Thank you for your replay;

Effectively, the 3D applications are often viewed as a temporal moving 2D slides; my principal goal is to define a theoretical approach to establish a higher dimension numerical convolutors.

Of course, the actual high performance computational architectures and the massive parallelism  are easily implemented on the real hardware or on the large computational grids as well as on the illimited virtual models.

My bests;

Sir,

Parallel processing is needed with Filters, Mesh and region growing Methodologies. for any data or image, 3D is possible by Simplex Meshes and by the conditions we apply for the growing the mesh is important here.

Prof Omar Bouattane, It is possible thru Meshes and Growing methods with specific conditions and parameters related to the objects which you are using to convert 2D to 3D.