def make_plots(results, img=None, label='', sidelabel=''):
ln = len(results)
if img is None:
size = 3
fig = pl.figure(figsize=(ln * size, size + 2.0 / ln))
img = ImageGrid(fig,
rect=[0.05 / ln, 0.05, 1 - 0.1 / ln, 0.90],
nrows_ncols=[1, ln],
axes_pad=0.1)
# get a common color bar scale for all images
mins, maxs = [], []
for i, v in enumerate(results):
if v[0] == 'Reference':
continue
mins.append(v[1].min())
maxs.append(v[1].max())
mins = min(mins)
maxs = max(maxs)
mins = -0.5 * max(np.abs([mins, maxs]))
maxs = -mins
for i, v in enumerate(results):
if v[0] == 'Reference':
img[i].imshow(v[1], vmin=0, vmax=1, cmap='bone')
else:
img[i].imshow(v[1], vmin=mins, vmax=maxs)
img[i].set_title(v[0], fontsize=17)
img[i].set_xticks([])
img[i].set_yticks([])
if label:
lbl(img[i], label + str(i + 1), 18)
if i == 0 and sidelabel:
img[i].set_ylabel(sidelabel)
def doplot(image0,
image1,
xs,
crbs,
errors,
labels,
diff_image_scale=0.1,
dolabels=True,
multiple_crbs=True,
xlim=None,
ylim=None,
highlight=None,
detailed_labels=False,
xlabel="",
title=""):
"""
Standardizing the plot format of the does_matter section. See any of the
accompaning files to see how to use this generalized plot.
image0 : ground true
image1 : difference image
xs : list of x values for the plots
crbs : list of lines of values of the crbs
errors : list of lines of errors
labels : legend labels for each curve
"""
if len(crbs) != len(errors) or len(crbs) != len(labels):
raise IndexError, "lengths are not consistent"
fig = pl.figure(figsize=(14, 7))
ax = fig.add_axes([0.43, 0.15, 0.52, 0.75])
gs = ImageGrid(fig,
rect=[0.05, 0.05, 0.25, 0.90],
nrows_ncols=(2, 1),
axes_pad=0.25,
cbar_location='right',
cbar_mode='each',
cbar_size='10%',
cbar_pad=0.04)
diffm = diff_image_scale * np.ceil(np.abs(image1).max() / diff_image_scale)
im0 = gs[0].imshow(image0, vmin=0, vmax=1, cmap='bone_r')
im1 = gs[1].imshow(image1, vmin=-diffm, vmax=diffm, cmap='RdBu')
cb0 = pl.colorbar(im0, cax=gs[0].cax, ticks=[0, 1])
cb1 = pl.colorbar(im1, cax=gs[1].cax, ticks=[-diffm, diffm])
cb0.ax.set_yticklabels(['0', '1'])
cb1.ax.set_yticklabels(['-%0.1f' % diffm, '%0.1f' % diffm])
image_names = ["Reference", "Difference"]
for i in xrange(2):
gs[i].set_xticks([])
gs[i].set_yticks([])
gs[i].set_ylabel(image_names[i])
if dolabels:
lbl(gs[i], figlbl[i])
symbols = ['o', '^', 'D', '>']
for i in xrange(len(labels)):
c = COLORS[i]
if multiple_crbs or i == 0:
if multiple_crbs:
label = labels[i] if (
i != 0 and not detailed_labels) else '%s CRB' % labels[i]
else:
label = 'CRB'
ax.plot(xs[i], crbs[i], '-', c=c, lw=3, label=label)
label = labels[i] if (
i != 0 and not detailed_labels) else '%s Error' % labels[i]
ax.plot(xs[i],
errors[i],
symbols[i],
ls='--',
lw=2,
c=c,
label=label,
ms=12)
if dolabels:
lbl(ax, 'D')
ax.loglog()
if xlim:
ax.set_xlim(xlim)
if ylim:
ax.set_ylim(ylim)
ax.legend(loc='upper left', ncol=2, prop={'size': 18}, numpoints=1)
ax.set_xlabel(xlabel)
ax.set_ylabel(r"Position CRB, Error")
ax.grid(False, which='both', axis='both')
ax.set_title(title)
return gs, ax
def plot_noise_pole(diff):
center = np.array(diff.shape) / 2
poles = np.array([
[190, 277],
[227, 253],
[233, 256],
[255, 511],
[255, 271],
#[255, 240],
#[198, 249],
#[247, 253]
])
q = np.fft.fftshift(np.fft.fftn(diff))
t = pole_removal(q, poles, sig=4)
r = np.real(np.fft.ifftn(np.fft.fftshift(t)))
fig = pl.figure(figsize=(24, 8))
gs = ImageGrid(fig,
rect=[0.05, 0.05, 0.33, 0.85],
nrows_ncols=(2, 2),
axes_pad=0.05)
images = [[diff, q], [r, t]]
maxs, mins = [], []
for i, im in enumerate(images):
a, b = im[0], np.abs(im[1])**0.2
maxs.append([a.max(), b.max()])
mins.append([a.min(), b.min()])
maxs, mins = np.array(maxs), np.array(mins)
maxs = 0.8 * maxs.max(axis=0)
mins = 0.8 * mins.min(axis=0)
labels = [['A', 'B'], ['C', 'D']]
for i, (im, lb) in enumerate(zip(images, labels)):
ax = [gs[2 * i + 0], gs[2 * i + 1]]
ax[0].imshow(im[0], vmin=mins[0], vmax=maxs[0], cmap=pl.cm.bone)
ax[1].imshow(np.abs(im[1])**0.2,
vmin=mins[1],
vmax=maxs[1],
cmap=pl.cm.bone)
lbl(ax[0], lb[0])
lbl(ax[1], lb[1])
for a in ax:
a.grid(False, which='both', axis='both')
a.set_xticks([])
a.set_yticks([])
if i == 0:
ax[0].set_title("Real-space")
ax[1].set_title("k-space")
ax[0].set_ylabel("Raw noise")
if i == 1:
ax[0].set_ylabel("Poles removed")
for p in poles:
for pp in [p, 2 * center - p]:
a.plot(pp[1], pp[0], 'wo')
ax = fig.add_axes([0.37, 0.15, 0.28, 0.7])
ax.imshow(diff - r, cmap=pl.cm.bone)
ax.set_xticks([])
ax.set_yticks([])
ax.set_title("Correction")
lbl(ax, 'E')
ax = fig.add_axes([0.70, 0.15, 0.25, 0.7])
sig = diff.std()
y, x = np.histogram(diff,
bins=np.linspace(-5 * sig, 5 * sig, 300),
normed=True)
x = (x[1:] + x[:-1]) / 2
ax.plot(x, y, '-', lw=2, alpha=0.6, label='Raw noise')
y, x = np.histogram(r,
bins=np.linspace(-5 * sig, 5 * sig, 300),
normed=True)
x = (x[1:] + x[:-1]) / 2
ax.plot(x, y, '-', lw=2, alpha=0.6, label='Poles removed')
ax.plot(x,
1 / np.sqrt(2 * np.pi * sig**2) * np.exp(-x**2 / (2 * sig**2)),
'k--',
lw=1.5,
label='Gaussian fit')
ax.set_xlabel("Pixel value")
ax.set_ylabel("Probability")
ax.legend(loc='best', prop={'size': 18})
ax.grid(False, which='minor', axis='y')
ax.locator_params(axis='x', nbins=5)
#ax.set_title("Distribution of residuals")
lbl(ax, 'F')
ax.semilogy()
def plot_noise(diff):
dat = diff.flatten()
sig = dat.std()
fig = pl.figure(figsize=(13, 4))
#=====================================================
gs2 = GridSpec(1,
2,
left=0.05,
bottom=0.22,
right=0.45,
top=0.85,
wspace=0.25,
hspace=0.25)
ax = pl.subplot(gs2[0, 0])
ax.imshow(diff[0], cmap=pl.cm.bone)
ax.set_xticks([])
ax.set_yticks([])
ax.grid(False, which='both', axis='both')
ax.set_xlabel(r"$x$")
ax.set_ylabel(r"$y$")
lbl(ax, 'A')
q = np.fft.fftn(diff)
q = np.fft.fftshift(np.abs(q[0])**0.1)
ax = pl.subplot(gs2[0, 1])
ax.imshow(q, cmap=pl.cm.bone)
ax.set_xticks([])
ax.set_yticks([])
ax.grid(False, which='both', axis='both')
ax.set_xlabel(r"$q_x$")
ax.set_ylabel(r"$q_y$")
lbl(ax, 'B')
#=====================================================
gs = GridSpec(1,
1,
left=0.59,
bottom=0.22,
right=0.89,
top=0.85,
wspace=0.35,
hspace=0.35)
y, x = np.histogram(dat,
bins=np.linspace(-5 * sig, 5 * sig, 300),
normed=True)
x = (x[1:] + x[:-1]) / 2
ax = pl.subplot(gs[0, 0])
ax.plot(x, y, '-', lw=2, alpha=0.6, label='Confocal image')
ax.plot(x,
1 / np.sqrt(2 * np.pi * sig**2) * np.exp(-x**2 / (2 * sig**2)),
'k--',
lw=1.5,
label='Gaussian fit')
ax.semilogy()
ymin = y[y > 0].min()
ymax = y[y > 0].max()
#ax.legend(loc='best', bbox_to_anchor=(0.75, 0.25), prop={'size':18})
ax.locator_params(axis='x', nbins=5)
ax.set_xlim(-5 * sig, 5 * sig)
ax.set_ylim(ymin, 1.35 * ymax)
ax.set_xlabel("Pixel value")
ax.set_ylabel("Probability")
ax.grid(False, which='minor', axis='y')
lbl(ax, 'C')
return gs
def smile_comparison_plot(state0, state1, stdfrac=0.7):
fig = pl.figure(figsize=(10, 12))
ig = ImageGrid(fig,
rect=[0.05, 0.55, 0.90, 0.40],
nrows_ncols=(1, 2),
axes_pad=0.2)
ax0 = fig.add_axes([0.13, 0.05, 0.74, 0.40])
ax1 = ax0.twinx()
stringer = lambda o: ', '.join([str(i) for i in o])
states = [state0, state1]
orders = [stringer(s.ilm.order) for s in states]
colors = ('#598FEF', '#BCE85B', '#6C0514')
dmin, dmax = 1e10, -1e10
rstd = -1e10
for i, (s, o, color) in enumerate(zip(states, orders, colors)):
ax = ig[i]
sl = np.s_[s.pad:-s.pad, s.pad:-s.pad, s.pad:-s.pad]
diff = -(s.image - s.get_model_image())[sl]
m = good_particles(s, False)
r = s.obj.rad[m]
std = stdfrac * r.std()
rstd = max([rstd, std])
dmin = min([dmin, diff.min()])
dmax = max([dmax, diff.max()])
for i, (s, o, color) in enumerate(zip(states, orders, colors)):
ax = ig[i]
m = good_particles(s, False)
p = s.obj.pos[m]
r = s.obj.rad[m]
z, y, x = p.T
sl = np.s_[s.pad:-s.pad, s.pad:-s.pad, s.pad:-s.pad]
mu = r.mean()
c = pl.cm.RdBu_r(Normalize(vmin=mu - rstd, vmax=mu + rstd)(r))[:, :3]
diff = -(s.image - s.get_model_image())[sl]
if i == 0:
lbl(ax, 'A')
elif i == 1:
lbl(ax, 'B')
ax.set_title(o)
ax.imshow(diff[-5], vmin=dmin, vmax=dmax)
ax.scatter(x - s.pad, y - s.pad, s=60, c=c)
ax.set_xlim(0, diff.shape[1])
ax.set_ylim(0, diff.shape[2])
ax.set_xticks([])
ax.set_yticks([])
ax.grid('off')
sect = 255 * nd.gaussian_filter(
diff.mean(axis=(0, 1)), 3, mode='reflect')
ax0.plot(sect, lw=2, c=color, label='Residual %s' % o)
ax1.plot(x, r, 'o', mfc=color, mec='black', label='Radii %s' % o, ms=6)
ax0.set_xlim(50, diff.shape[1])
ax1.set_xlim(50, diff.shape[1])
ax0.set_ylim(sect.mean() - 16 * sect.std(),
sect.mean() + 26 * sect.std())
ax1.set_ylim(r.mean() - 5 * r.std(), r.mean() + 2 * r.std())
ax0.grid('off')
ax1.grid('off')
ax0.set_xticks([])
ax1.set_xticks([])
ax0.set_xlabel("Pixel position")
ax0.set_ylabel("Residual value")
ax1.set_ylabel("Particle Radius")
ax0.legend(bbox_to_anchor=(1.08, 1.3), ncol=4)
ax1.legend(bbox_to_anchor=(1, 1.17), ncol=4, numpoints=1)
if i == 1:
lbl(ax0, 'C')
lbl(ax1, 'C')