Reducing the effects of noise in image reconstruction (Q1610524)

The authors continue their investigation on image reconstruction from Fourier spectral data based in particular on the so-called concentration method, as developed by them. This method requires a priori knowledge of the jump discontinuity locations and amplitude when the image is viewed as a piecewise smooth function. The authors describe the evolution of the concentration method and combine it with the Gegenbauer reconstruction method. As random noise may exist in the spectral data, they investigate the ability to locate the edges and to recover the image in the presence of noise. Their results are illustrated with examples of the Shepp Logan brain phantom.