Wavelets on irregular point sets
Ingrid C. Daubechies, Igor Guskov, Peter Schröder, Wim Sweldens · 2000
Abstract Wavelets are a versatile tool for representing general functions and datasets, and they enjoy widespread use in areas as diverse as signal processing, image compression, finite-element methods and statistical analysis (among many others). In essence we may think of wavelets as building blocks with which to represent data and functions. The particular appeal of wavelets derives from their representational and computational efficiency: most datasets exhibit correlation both in time (space) and frequency, as well as other types of structure. These can be modelled with high accuracy through sparse combinations of wavelets. Wavelet representations can also be computed fast, because they can be built using multiresolution analysis and subdivision.