In [1]:
%pylab inline
from skimage import (
    io as dataio, 
    feature, 
    filters, 
    measure, 
    morphology, 
)

Populating the interactive namespace from numpy and matplotlib

Opening the image

In [2]:
image = dataio.imread('https://imagej.nih.gov/ij/images/blobs.gif')
imshow(image, cmap='gray')
Out[2]:
<matplotlib.image.AxesImage at 0x7f64837f9a58>

Duplicate & renaming window not needed 😉, Moving straight to blurring!

In [3]:
blurred = filters.gaussian(image, sigma=5)
imshow(blurred, cmap='gray')
Out[3]:
<matplotlib.image.AxesImage at 0x7f64832d3208>

Finding maxima (and showing on plot)

Note:

  • Instead of specifying "Light background" which means finding negative peaks, we simply find peaks in the negative image more explicitly
  • By default peaks are returned in a 2d list of x,y points.
In [4]:
maxima = feature.peak_local_max(-blurred, min_distance=10)
imshow(blurred, cmap='gray')
plot(maxima[:, 1], maxima[:, 0], 'y+')
Out[4]:
[<matplotlib.lines.Line2D at 0x7f6483301b38>]

Analysis

In [11]:
maxima_image = feature.peak_local_max(-blurred, min_distance=10, indices=False)
props_table = measure.regionprops_table(
    measure.label(maxima_image), image, 
    properties=['centroid', 'max_intensity'])
import pandas as pd
pd.DataFrame(props_table)
Out[11]:
centroid-0 centroid-1 max_intensity
0 15 19 31.0
1 15 110 7.0
2 16 197 7.0
3 26 44 23.0
4 28 218 87.0
5 32 173 39.0
6 41 70 47.0
7 47 233 7.0
8 54 139 7.0
9 61 27 47.0
10 65 192 39.0
11 71 214 23.0
12 73 167 55.0
13 74 98 23.0
14 83 63 39.0
15 85 126 7.0
16 86 33 31.0
17 95 222 23.0
18 99 176 23.0
19 111 103 39.0
20 122 37 39.0
21 123 62 39.0
22 124 130 55.0
23 127 222 7.0
24 138 168 15.0
25 139 122 39.0
26 146 67 23.0
27 157 188 47.0
28 163 136 39.0
29 166 214 55.0
30 170 16 39.0
31 172 44 31.0
32 174 86 7.0
33 180 180 23.0
34 185 233 7.0
35 197 129 47.0
36 203 206 71.0
37 209 24 55.0
38 214 103 7.0
39 218 62 55.0
40 228 234 7.0
41 229 178 23.0
42 233 138 47.0
43 241 46 31.0
In [ ]: