def _runImp(self):
if self._N0_32F is None:
self._computeInitialNormal()
self._optimize()
L = self._L
self._I_32F = diffuse(self._N_32F, L)
reflectance = LambertReflectanceEstimation(self._C0_32F, self._I_32F)
self._C_32F = reflectance.shading(self._I_32F)
def _runImp(self):
if self._N0_32F is None:
self._computeInitialNormal()
self._optimize()
L = self._L
self._I_32F = diffuse(self._N_32F, L)
reflectance = LambertReflectanceEstimation(self._C0_32F, self._I_32F)
self._C_32F = reflectance.shading(self._I_32F)
def errorTable():
colormap_files = colorMapFiles()
num_colormap_files = len(colormap_files)
M_errors = np.zeros((num_colormap_files))
for mi, colormap_file in enumerate(colormap_files):
M_32F = loadColorMap(colormap_file)
C_32F = M_32F.reshape(1, len(M_32F), 3)
I_32F = np.linspace(0.0, 1.0, len(M_32F))
I_32F = I_32F.reshape(C_32F.shape[:2])
reflectance = LambertReflectanceEstimation(C_32F, I_32F)
Ml = reflectance.shading(I_32F)[0, :, :]
I0_32F = luminance(C_32F)
IL_32F = luminance(Ml.reshape(1, -1, 3))
I_min, I_max = np.min(I0_32F), np.max(I0_32F)
M_error = normVectors(M_32F - Ml)
#M_errors[mi] = np.mean(M_error) / (I_max - I_min)
# M_errors[mi] = computeGradientDistance(M_32F, Ml) / (I_max - I_min)
#M_errors[mi] = np.linalg.norm(I0_32F - IL_32F) / (I_max - I_min)
M_errors[mi] = np.mean(M_error) / (I_max - I_min)
# M_errors[mi] = np.linalg.norm(M_32F - Ml) / (I_max - I_min)
# M_errors[mi] = compareHist(M_32F, Ml)
file_path = shapeResultFile("ReflectanceEstimation", "ReflectanceError", file_ext=".npy")
np.save(file_path, M_errors)
plt.plot(M_errors)
plt.show()
def reflectanceEstimationFigure():
errorTable()
M_errors = loadReflectanceErrorTable()
M_error_orders = np.argsort(M_errors)
print M_errors[M_error_orders]
colormap_files = colorMapFiles()
colormap_files = [colormap_files[M_error_order] for M_error_order in M_error_orders]
colormap_files = colormap_files[0:-1:3]
Ms = []
MLs = []
for colormap_file in colormap_files:
M_32F = loadColorMap(colormap_file)
Ms.append(M_32F)
C_32F = M_32F.reshape(1, len(M_32F), 3)
I_32F = np.linspace(0.0, 1.0, len(M_32F))
I_32F = I_32F.reshape(C_32F.shape[:2])
reflectance = LambertReflectanceEstimation(C_32F, I_32F)
Ml = reflectance.shading(I_32F)
MLs.append(Ml[0, :, :])
num_rows = 3
num_cols = len(colormap_files)
w = 10
h = w * num_rows / num_cols
fig, axes = plt.subplots(figsize=(w, h))
font_size = 15
fig.subplots_adjust(left=0.02, right=0.98, top=0.98, bottom=0.02, hspace=0.05, wspace=0.05)
fig.suptitle("", fontsize=font_size)
N_sphere, A_32F = normalSphere(h=512, w=512)
Lg = normalizeVector(np.array([-0.2, 0.3, 0.6]))
plot_grid = SubplotGrid(num_rows, num_cols)
mi = 1
for M, Ml in zip(Ms, MLs):
CM_32F = ColorMapShader(M).diffuseShading(Lg, N_sphere)
CL_32F = ColorMapShader(Ml).diffuseShading(Lg, N_sphere)
C_error = normVectors((CM_32F - CL_32F).reshape(-1, 3)).reshape(CL_32F.shape[:2])
C_error[A_32F < 0.5 * np.max(A_32F)] = 0.0
plot_grid.setPlot(1, mi)
plot_grid.showImage(setAlpha(CM_32F, A_32F), "")
plot_grid.setPlot(2, mi)
plot_grid.showImage(setAlpha(CL_32F, A_32F), "")
plot_grid.setPlot(3, mi)
plot_grid.showColorMap(C_error, "", v_min=0.0, v_max=0.3, with_colorbar=False)
mi += 1
file_path = shapeResultFile("ReflectanceEstimation", "ReflectanceEstimationError")
fig.savefig(file_path, transparent=True)