engleski

Video Denoising Using the ICI Method

In this paper we have proposed a method for video denoising using the intersection of confidence intervals (ICI) rule. Among various noise removing techniques in both frequency and time domains, the adaptive method based on the ICI rule has many useful features, one for example being an efficient edge preserving. Due to the filter adaptivity, denoising performed using the here proposed method results in noise suppression of video segments recording stationary scenes, as well as video segments with objects in motion. It is shown that this adaptive algorithm is superior to fixed size filters based algorithms, which are suitable only for sequences of images recording stationary scenes. The method was tested on two real-life videos for a range of noise levels, achieving the peak signal-to-noise ratio (PSNR) improvement by up to 23 dB when compared to the best fixed size (not known in advance) filter based approach.

Intersection of confidence intervals (ICI) rule; video denoising; image denoising; adaptive filtering; edge preserving

nije evidentirano