Signal smoothness estimation in hölder spaces
Mong-Shu Lee · International Journal of Computer Mathematics · 2000
Previous research has shown that wavelet method can be used to estimate the Besov smoothness of a function (signal). This paper describes an algorithm that is based on the magnitudes of the wavelet coefficients and linear regression model to estimate the smoothness of different signals of one and two-dimensional in the Hölder spaces. Computational results show that the Holder smoothness of the general two-dimensional image is between 0.2 and 0.7. We compare our results with those in Besov smoothness spaces and discuss the smoothness relations between these two function spaces.