期刊名称：International Journal of Signal Processing, Image Processing and Pattern Recognition
印刷版ISSN：2005-4254
出版年度：2016
卷号：9
期号：9
页码：401-418
出版社：SERSC
摘要：This paper proposes a hybrid technique to remove impulse noise from digital images. In this approach the filtering operation is based on 33 neighborhood of pixel under consideration. During filtering, the properties of neighborhood are considered to check whether it is highly corrupted with noise, medium or only itself act as impulse. Based upon these properties a new hybrid technique has been proposed to process the pixel which further uses different schemes. The experiments have been performed at various noise levels on standard images as well as on real images. The results have been evaluated on the basis of metrics like Signal to noise ratio (SNR), Edge preservation index( EPI), Structure similarity index measure (SSIM), Multi scale structure similarity index measure (MS-SSIM) and Peak signal to noise ratio( PSNR). From the results, ithas been observed that proposed technique has worked efficiently by preserving the edges and fine lines. To demonstrate the effectiveness of proposed technique, the results have also been compared with other well accepted denoising techniques
关键词：Impulse noise; Noise filtering; Median filter; Midpoint filter; Image ;restoration
