Robust Noise Filtering in Image Sequences
This addresses noise filtering for multimedia and video systems, but appears incremental as it builds on anisotropic diffusion techniques.
The authors tackled the problem of noise removal in image sequences, which are often corrupted by sensor noise, by proposing a robust method based on coupled spatial and temporal anisotropic diffusion, achieving adaptive smoothing while preserving discontinuities.
Image sequences filtering have recently become a very important technical problem especially with the advent of new technology in multimedia and video systems applications. Often image sequences are corrupted by some amount of noise introduced by the image sensor and therefore inherently present in the imaging process. The main problem in the image sequences is how to deal with spatio-temporal and non stationary signals. In this paper, we propose a robust method for noise removal of image sequence based on coupled spatial and temporal anisotropic diffusion. The idea is to achieve an adaptive smoothing in both spatial and temporal directions, by solving a nonlinear diffusion equation. This allows removing noise while preserving all spatial and temporal discontinuities