Dynamic restoration of periodic noisy image using: Upper-half spectrum

This paper develops a new approach of image restoration from multi periodic noises. It utilized the property of the conjugate symmetry in frequency domain to clean just a half-size of image band. First, the auto detection and recovery procedure was applied to remove or decrease the effect of the degradation from a whole frequency domain image by locating the spikes and replace them by zero. The procedure is adapted to clean just a half image size, or the upper half-spectrum of image band. The mirror property of the two dimensions discrete Fourier transform is used to re-obtain the lower image band, where a re-allocating conjugate symmetry procedure is applied to refill the deleted locations by new de-noisy bands. A comparison between the two restoration methods is made by measuring the qualification using peak signal to noise ratio. Saving in time and space could be satisfied by processing just the upper-half of image band.