Sharpening an image

This notebook sharpens an AFM image in two ways, on the height data themselves. Sharpening is for looking at fine detail: it changes the heights, so make your measurements on the unsharpened image. Run it from this folder, next to grid.nid.

from NSFopen.process import process

import matplotlib.pyplot as plt
from scipy.ndimage import laplace

afm = process("grid.nid", verbose=False)
afm.flatten(order=1)
z = afm.array()

Unsharp mask

An unsharp mask adds the difference between the image and a blurred copy of it. sigma is the blur in pixels and amount the strength.

Laplacian

Subtracting the Laplacian, the local curvature of the image, sharpens the finest detail, with amount as its strength.

set_channel stores each result as a channel of its own, so the plots and the analysis can use them like any other.

unsharp = afm.filter_basic(inplace=False, kind="sharpen", sigma=2.0, amount=3.0)
afm.set_channel(unsharp, channel="Unsharp mask", unit="m")

afm.set_channel(z - 2.0 * laplace(z), channel="Laplacian", unit="m")

fig, axes = plt.subplots(1, 3, figsize=(14, 4.2), layout="constrained", dpi=150)
for ax, channel in zip(axes, ("Z-Axis", "Unsharp mask", "Laplacian")):
    afm.image(ax=ax, channel=channel, color_range="p1")
plt.show()

A line cut across all three shows what happens to the heights: edges get steeper, and the unsharp mask overshoots beside them.

width, height = afm.size()
ends = (0.1 * width, 0.5 * height), (0.6 * width, 0.5 * height)

fig, ax = plt.subplots(figsize=(7, 4), layout="constrained", dpi=150)
for channel in ("Z-Axis", "Unsharp mask", "Laplacian"):
    profile = afm.line_profile(*ends, channel=channel)
    ax.plot(profile.distance * 1e6, profile.mean * 1e9, label=channel, linewidth=1)
ax.set(xlabel="Distance [µm]", ylabel="Height [nm]")
ax.legend(frameon=False)
plt.show()