Fast rank algorithms based on multiscale histograms and lazy calculations
- 125 Downloads
Rank algorithms allow effective solutions for image smoothing and impulse noise suppression, but most of them are computationally complex. On the base of multiscale histograms, we propose algorithms for fast computations of EV and KNV neighborhood average, sliding equalization and search for an arbitrary element in a rank series (median filtering is a particular case of this algorithm). An approach using lazy calculations for fast updating of multiscale histograms is proposed. Using the developed algorithms, we have achieved a processing speed for EV and KNV neighborhood average algorithms that is close to the fastest known median filtering algorithm.
Keywordsrank algorithm image denoising image sharpening image filtering lazy calculations
Unable to display preview. Download preview PDF.
- 3.M. V. Storozhilova and D. V. Yurin, “Fast rank algorithms based on multiscale histograms,” in Proc. 21st Int. Conf. on Computer Graphics GraphiCon’2011 (Moscow, 2011), pp. 132–135.Google Scholar
- 4.C. Tomasi and R. Manduchi, “Bilateral filtering for gray and color images,” in Proc. 6th Int. IEEE Conf. on Computer Vision (ICCV’98) (Bombay, Jan. 1998), pp. 839–846.Google Scholar
- 7.M. Storozhilova and D. Yurin, “Fast rank algorithms with multiscale histograms lazy updating,” in Proc. 8th Open German-Russian Workshop “Pattern Recognition and Image Understanding” (OGRW-8-2011) (Nizhni Novgorod, Nov. 2011), pp. 380–383.Google Scholar