Figure 5
Implementation-adapted filters for noisy and sparsely sampled data. (Top, left to right) Mean standard deviations for slice S = 128 as a function of the number of projection angles Nθ, mean value of the squared bias, mean value of RMSE with respect to the ground truth slice, and optimized filters in Fourier space. (Bottom, left to right) Mean standard deviations in S = 128 as a function of photon flux I0 (higher values of I0 correspond to lower noise levels) using Nθ = 64, mean value of the squared bias, mean value of RMSE with respect to the ground truth slice, and optimized filters in Fourier space. |