Edge-preserving regularized filter with visual perception outlier measurement

Abstract

In this paper, we propose an impulse noise removal algorithm, which can realize the elimination of the impulse noise in an image effectively even though these noises are on image edges or details. The algorithm introduces a statistic, visual perception outlier measurement, to identify the impulse noise in an image. Once the noises are identified, noise candidates are restored by a nonlocal edge-preserving regularized filter. Extensive experimental results indicate that our algorithm provides a significant improvement over many other existing techniques.

Keywords:

• Impulse noise
• visual perception outlier measurement
• edge-preserving regularized (EPR)