EFFICIENT AND RELIABLE SCHEMES FOR NONLINEAR DIFFUSION FILTERING

Nonlinear diffusion filtering is usually performed with explicit schem es, They are only stable for very small time steps, which leads to poo r efficiency and limits their practical use, Based on a recent discret e nonlinear diffusion scale-space framework we present semi-implicit s chemes which are stable for all time steps, These novel schemes use an additive operator splitting (AOS), which guarantees equal treatment o f all coordinate axes. They can be implemented easily in arbitrary dim ensions, have good rotational invariance and reveal a computational co mplexity and memory requirement which is linear in the number of pixel s. Examples demonstrate that, under typical accuracy requirements, AOS schemes are at least ten times more efficient than the widely used ex plicit schemes.