def mat_comp():
tik_alpha = 1
tv_alpha = 0.1
# Creating measurement data
data = r.radon(phantom[n], theta)
A = r.radon_matrix(n_ang, n)
At = np.transpose(A)
# Adding noise
noise_level = 0.05
noisy_data = data + noise_level * np.random.randn(data.shape[0],
data.shape[1])
# Compute the reconstruction
rec_tik = tik_primal_dual_mat_alt(A, At, noisy_data, (n, n), tik_alpha)
rec_ls = LS_primal_dual_mat(A, At, noisy_data, (n, n))
rec_tv = TV_primal_dual_mat(A, At, noisy_data, (n, n), tv_alpha)
pylab.gray()
pylab.figure(0)
pylab.title("Least-Squares reconstruction")
pylab.imshow(rec_ls)
pylab.figure(1)
pylab.title("Tikhonov reconstruction")
pylab.imshow(rec_tik)
pylab.figure(2)
pylab.title("Total Variation reconstruction")
pylab.imshow(rec_tv)
pylab.show()
def comp():
tik_alpha = 1
# Creating measurement data
data = r.radon(phantom[n], theta)
# Adding noise
noise_level = 0.05
noisy_data = data + noise_level * np.random.randn(data.shape[0],
data.shape[1])
# Compute the reconstruction
rec_tik = tik_primal_dual_alt(noisy_data, (n, n), tik_alpha)
rec_ls = LS_primal_dual(noisy_data, (n, n))
pylab.gray()
pylab.figure(0)
pylab.title("Least-Squares reconstruction")
pylab.imshow(rec_ls)
pylab.figure(1)
pylab.title("Tikhonov reconstruction")
pylab.imshow(rec_tik)
pylab.show()
if show_prog:
if i % 50 == 0:
gap = 1/2*np.linalg.norm(A.dot(ft)-meas_v, 2)**2 + 1/2 * \
np.linalg.norm(pk, 2)**2 + np.dot(pk, meas_v)
pylab.figure(2)
pylab.title("Iteration {} | Dual gap: {:.2f}".format(i, gap))
pylab.gray()
pylab.imshow(np.reshape(ft, f_shape))
pylab.pause(0.1)
return np.reshape(ft, f_shape)
if __name__ == "__main__":
# Creating measurement data
data = r.radon(phantom[n], theta)
#A = r.radon_matrix(theta, n)
#At = np.transpose(A)
# Adding noise
noise_level = 0.05
noisy_data = data + noise_level * np.random.randn(data.shape[0],
data.shape[1])
# Compute the reconstruction
rec = LS_primal_dual(noisy_data, (n, n), True)
#rec = LS_primal_dual_mat(A, At, noisy_data, (n,n), True)
pylab.gray()
pylab.figure(0)
def Af(f):
# Define the forward projection with given number of angles
global theta
return r.radon(f, theta)
def A(f):
global theta
return r.radon(f, theta)