n : float
nu value
l : float
lambda value
Returns
-------
B : numpy.ndarray
The 3 x P matrix of pesudonormals
"""
g = np.eye(3)
g[2, 0] = m
g[2, 1] = n
g[2, 2] = l
B = np.dot(np.linalg.inv(g).T, B)
return B
if __name__ == "__main__":
I, L_0, s = loadData()
B, L = estimatePseudonormalsUncalibrated(I)
B = enforceIntegrability(B, s, 3)
B = bas_relief(B, 0, 0, 2)
albedos, normals = estimateAlbedosNormals(B)
albedoIm, normalIm = displayAlbedosNormals(albedos, normals, s)
print(L)
print(L_0)
surface = estimateShape(normals, s)
plotSurface(surface)
I, realL, s = I, L, s = loadData()
B, L = estimatePseudonormalsUncalibrated(I)
print(L)
print(realL)
_, normalsBad = estimateAlbedosNormals(B)
BI = enforceIntegrability(B, s)
# albedos, normals = estimateAlbedosNormals(BI)
# displayAlbedosNormals(albedos, normals, s)
# surf1 = estimateShape(normalsBad, s)
# surf2 = estimateShape(normals, s)
# plotSurface(np.clip(surf1, -1000, 1000))
# plotSurface(surf2)
G = np.eye(3)
G[2, 2] = 0
G[2, :] = [10, 0, .7]
Ginv = np.linalg.inv(G)
Bnew = Ginv.T @ BI
_, normals = estimateAlbedosNormals(Bnew)
surf = estimateShape(normals, s)
plotSurface(surf)
albedos_img, normals_img = displayAlbedosNormals(albedos, normals, shapes)
for itr in range(6):
lightning_estimated[itr, :] = lightning_estimated[
itr, :] / np.linalg.norm(lightning_estimated, axis=1)[itr]
print("ground truth lighting\n", lighting.T)
print("estimated lightning\n", lightning_estimated)
#2d
normals_reshape = np.zeros((3, shapes[0] * shapes[1]))
normals_reshape[0, :] = normals_img[:, :, 0].reshape(-1)
normals_reshape[1, :] = normals_img[:, :, 1].reshape(-1)
normals_reshape[2, :] = normals_img[:, :, 2].reshape(-1)
surface_1 = estimateShape(normals_reshape, shapes)
plotSurface(surface_1)
#2e
enforced_pseudo_normals = enforceIntegrability(pseudo_normals, shapes)
lamb = 1
v = 0.01 #increases v will decrease the difference along lateral axis
mu = 0.01 #increase mu will decrease the depth difference among the faces
G = np.asarray([[1, 0, 0], [0, 1, 0], [mu, v, lamb]])
enforced_pseudo_normals = G.T @ enforced_pseudo_normals
albedos, normals = estimateAlbedosNormals(enforced_pseudo_normals)
albedos_img, normals_img = displayAlbedosNormals(albedos, normals, shapes)
normals_reshape = np.zeros((3, shapes[0] * shapes[1]))
plt.show() plt.figure() plt.imshow(normnormalIm, cmap = 'rainbow') plt.show() #2c plt.figure() plt.imshow(L, cmap = 'hot') plt.figure() plt.imshow(Lest, cmap = 'hot') plt.show() #2d surface = estimateShape(normals, s) plotSurface(surface) # #2e Bt = enforceIntegrability(B, s) albedost, normalst = estimateAlbedosNormals(Bt) surfacet = estimateShape(normalst, s) # plotSurface(surfacet) #2f #Bas Relief mu = 0 v = 0 l = 0.000001 G = np.array([[1,0,0],[0,1,0],[mu,v,l]])