期刊名称：International Journal of Computer Science & Technology
印刷版ISSN：2229-4333
电子版ISSN：0976-8491
出版年度：2013
卷号：4
期号：2
页码：465-468
语种：English
出版社：Ayushmaan Technologies
摘要：Image filtering algorithms are applied on images to remove the different types of noise that are either present in the image during capturing or injected in to the image during transmission. Digital images when captured usually have Gaussian noise, Speckle noise and salt and pepper noise [10]. The performances of the filters are compared using the Peak Signal to Noise Ratio (PSNR) and Mean Square Error (MSE) [8-9]. In this paper we proposed a spatial domain filter based on pattern recognition technique [1] using varying size mask not necessarily square matrix but ranging from 4 x 4 to 7 x 7 size mask .In the proposed filter the MSE is reduced to 43 % in case of saltandpepper noise, 51 % for Gaussian noise, 52 % for speckle noise and PSNR is increased to 6 % in case of saltandpepper noise, 12 % for Gaussian noise, 12 % for speckle noise. The image used for the purpose of analysis is standard woman hat image taken by Ron Bucci.
关键词：Salt and Pepper Noise;Gaussian Noise;Speckle Noise;Pattern Recognition Filter;PSNR;MSE
