Adaptive image smoothing algorithms for edge and texture preservation
Keda Tao · 2005
A class of reasonably simple and efficient adaptive algorithms are developed to enhance noise degraded images while preserving the edge and texture information. Techniques from one-dimensional adaptive signal processing and systems identification are extended and applied to two-dimensional image smoothing through proper modelling of the image. Both the AR and ARMA models are treated. The conceptual separation of "image causality" and "processing causality" is advocated. Advanced topics that are covered include: faster computation algorithms, simultaneous contrast stretching and smoothing, etc.