Removal of impulse noise using gain factors adapted by noise-free pixel number and pixel variation

Ching Ta Lu, Jun Hong Shen, Mu Yen Chen, Ling Ling Wang, Chih Chan Hsu

Research output: Chapter in Book/Report/Conference proceedingConference contribution


Impulse noise impacts an image, causing the quality of image to be deteriorated in image transmission or capture. In this paper, we propose a gain factor for the removal of the impulse noise. A 3 × 3 fixed-size local window is employed to analyze each extreme pixel (0 or 255 for an 8-bit gray-level image). All non-extreme pixels are sorted in an ascending order and are grouped according to the variation of pixel levels. If the pixel level between adjacent two sorted pixels varies seriously, a new group is created. Hence, the ratio and median value of each group are computed to determine the values of the gain factors. They are multiplied with the median value of each group to obtain the weighted value which is employed to replace the center pixel with an extreme value, enabling noise-corrupted pixels to be restored. Experimental results show that the proposed method can effectively remove salt-and-pepper noise from a corrupted image for various noise corruption densities (from 10% to 90%); meanwhile, the denoised image is freed from the blurred effect.

Original languageEnglish
Title of host publicationFrontier Computing - Theory, Technologies and Applications FC 2017
EditorsNeil Y. Yen, Jason C. Hung, Lin Hui
PublisherSpringer Verlag
Number of pages11
ISBN (Print)9789811073977
Publication statusPublished - 2018
Event6th International Conference on Frontier Computing, FC 2017 - Osaka, Japan
Duration: 2017 Jul 122017 Jul 14

Publication series

NameLecture Notes in Electrical Engineering
ISSN (Print)1876-1100
ISSN (Electronic)1876-1119


Conference6th International Conference on Frontier Computing, FC 2017

All Science Journal Classification (ASJC) codes

  • Industrial and Manufacturing Engineering


Dive into the research topics of 'Removal of impulse noise using gain factors adapted by noise-free pixel number and pixel variation'. Together they form a unique fingerprint.

Cite this