Variational Regularization of 3D Data provides an introduction to variational methods for data modelling and its application in computer vision. In this book, the authors identify interpolation as an inverse problem that can be solved by Tikhonov regularization. The proposed solutions are generalizations of one-dimensional splines, applicable to n-dimensional data and the central idea is that these splines can be obtained by regularization theory using a trade-off between the fidelity of the data and smoothness properties.

As a foundation, the authors present a comprehensive guide to the necessary fundamentals of functional analysis and variational calculus, as well as splines. The implementation and numerical experiments are illustrated using MATLAB®. The book also includes the necessary theoretical background for approximation methods and some details of the computer implementation of the algorithms. A working knowledge of multivariable calculus and basic vector and matrix methods should serve as an adequate prerequisite.

Includes supplementary material: sn.pub/extras

Autorentext

Hebert Montegranario is an applied mathematician and computational scientist, currently serving as a professor at the Antioquia University, Medellín, Colombia. He holds a Ph.D. in Engineering from Universidad Nacional de Colombia, with a research focus on image processing, computer vision, and machine learning. Dr. Montegranario has extensive expertise in numerical analysis, scientific computing, and software design. He has authored several academic papers and a book on variational regularization of 3D data, showcasing his contributions to the field.

Inhalt

3D Data in Computer vision and technology.- Function Spaces and Reconstruction.- Variational methods.- Interpolation: From one to several variables.- Functionals and their physical interpretations.- Regularization and inverse theory.- 3D Interpolation and approximation.- Radial basis functions.