Computational methods for modelling of nonlinear systems (Q879244)

The present book is concerned with nonlinear systems which can be described from observations of their input-output signals only. A nonlinear system is understood in a broad sense, as a device which transforms signals. The authors present in the first part the operator theoretic background, centered around the basic ideas of approximation, numerical methods, mathematical statistics and optimal filtering of stochastic signals. In the second part, they describe specific computational methods which follow from the general framework, mainly on filtering and image processing. The book contains a lot of examples of processed digital pictures. The book is divided into two parts. Part I, Methods of operator approximation in system modelling, has the following chapters: 2. Nonlinear operator approximation with preassigned accuracy, 3. Interpolation of nonlinear operators, 4.Realistic operators and their approximation, 5. Methods of best approximation for nonlinear operators. The second part, Optimal estimation of random vectors, contains the following chapters: 6. Computational methods for optimal filtering of stochastic signals, and 7. Computational methods for optimal compression and reconstruction of data. The first chapter of the book, Overview, contains a general presentation of the developed topics. Reporting on recent developments in systems theory, with a rigorous theoretical foundation, many examples and open problems, the book will be of interest for researchers who are using operator theoretic methods in nonlinear systems, filtering, image processing, data compression and reconstruction.