Periodic noise recognition and elimination using RFPCM clustering

Frequency domain filtering is a conventional method in eliminating periodic noise in an image. Conventional methods include filtering via bandpass filter and notch filter for detecting both periodic and quasi-periodic noise in an image. In this paper, RFPCM clustering has been used to detect periodic noise spikes in Fourier amplitude domain. Qualitative and comparative analysis has been presented in this paper for clustered mask method and the conventional method as well. PSNR values of the resultant images are taken as filtering quality factor and comparison analysis is taken with respect to this value.