Decomposition of stochastic properties within images using non-parametric methods

This paper discusses the application of three different nonparametric methods for decomposing images into regions which exhibit special stochastic properties, together with the constituent components of the stochastic. These are: (1) order statistics in connection with steps in the empirical estimated distribution functions; (2) detection of stochastic information within an image by hypothesis testing; and (3) rank order statistics to decompose the different types of stochastic within an image. The decomposition is used to isolate different image regions and to estimate the processes which are the constituent stochastic components. In order to achieve this, decisions based upon membership relations are employed and adapted thresholds used. The thresholds are obtained by the ordering of terms calculated by stochastic estimation methods together with one of the aforementioned techniques