Auto detection and removal of frequency domain periodic noise
Muthana H. Hamd · 2014
Periodic noise has a global effect, so it cannot be removed and hence recovered in data or spatial domain. This paper develops two approaches of auto detection and recovery of two spikes and their associated degradation lines using frequency domain edge detection. These procedures apply two separated band reject filters to avoid blocking the unaffected bands. Also, the Ring and Notch filters have been modified to detect and recover automatically the degradation bands by applying an approximation value instead of zeros. The locations of the two spikes and their associated lines in frequency domain reflect the periodic noise frequency in spatial domain, so the process of detection and recovery is implemented only in the two quarters of the transformed Red, Green, and Blue planes. Those methods have been implemented and compared using Peak Signal to Noise Ratio quality measurement. The result shows that the developed methods achieved better decibel quality over Ring and Notch method; they are suitable for rapid conducting of maximum qualification and hence a constant steady state vs. noise frequency relation.