Dissection capability analysis of noise suppression technique utilizing trilateral filter under mixed noise circumstances
Vorapoj Patanavijit, Kornkamol Thakulsukanant · 2017
Since 1998, the Bilateral filter (BF) has been proposed for suppressing Gaussian noise however the BF cannot functionally perform under the mix noise (that is comprised of Gaussian and S&P noise). Thereby, the Trilateral filter (TF) that is an adapted BF formed on the ROAD (Rank-Ordered Absolute Differences) technique was initially stated by Roman Garnett et al. for self-regulating suppressing and eliminating of the mix noise in 2005. This paper analytically investigates the dissection capability consequence of 4 parameters (joint impulsivity variance, ROAD variance, spatial variance and radiometric variance) of the TF for suppressing several mixed noise circumstances. In this paper, five contaminated images (Lena, Girl-Tiffany, Mobile, Baboon, House) under many mixed Gaussian and S&P noise circumstances are used to investigate the maximum TF capability in the PSNR aspect, contemplated with the ordinary techniques: median filter, mean filter and BF (BF). Later, the dissection analytical value of joint impulsivity variance and ROAD variance that make the highest PSNR is investigated for each image and each mix noise case.