Compressive sampling on the Fourier domain as a method for increasing image signal-to-noise ratio

Abstract

Compressive sampling using measurements taken from a line mask on the Fourier domain is used to reconstruct test images with Gaussian, speckle, 1/f, and uniform noise. Using signal-to-noise ratio (SNR), and Linfoot’s criteria as similarity measurements, we are able to show that the images with noise of varying strengths sampled at 5%, 10%, 20% and 30% of Nyquist criteria are generally better compared to completely sampled images.