A new adaptive switching filter with directional impulse detection

Abstract

A novel adaptive switching filter (ASF) based on directional detection is proposed for denoising the images that are highly corrupted by impulse noise. The proposed algorithm employs an efficient noise detection mechanism. It first employs an efficient method to estimate the differences between the current pixel and its neighbors aligned with 28 directions. The current noise pixel is replaced by a median or a mean value within an adaptive filter window with respect to different noise densities. Experimental results show that the proposed approach can not only achieve very low miss-detection ratio and false-alarm ratio even up to high noise corruption, but also preserve the detailed information of an image very well.

Keywords

Impulse Noise / Center Pixel / Noise Density / Current Pixel / Noise Pixel

References

[1] SunT, NeuvoY. Pattern Recognit. Lett., 1994, 15: 341

[2] HwangH, HaddadR A. IEEE Trans. Image Process., 1995, 4: 499

[3] GonzalezR C, WoodsR E. Digital Image Processing, 2002, 2nd ed.Upper Saddle River, NJ, Prentice Hall

[4] EngH L, MaK K. IEEE Trans. Image Process., 2001, 10: 242

[5] SrinivasanK S, EbenezerD. IEEE Signal Process. Lett., 2007, 14: 189

[6] DongY Q, XuS F. IEEE Signal Process. Lett., 2007, 14: 193

[7] LuC T, ChouT C. Pattern Recognit. Lett., 2012, 33: 1287

[8] NgP E, MaK K. IEEE Trans. Image Process., 2006, 15: 1506

[9] DuanF, ZhangY J. IEEE Signal Process. Lett., 2010, 17: 647

[10] PitasI, VenetsanopoulosA N. Nonlinear Digital Filter: Principles and Application, 1990, Boston, MA, Kluwer