Debian Codegraph @

doc/features/image/menu_analysis.rst

raw
1 .. _ima-menu-analysis:
2
3 Analysis features on Images
4 ============================
5
6 This section describes the image analysis features available in DataLab.
7
8 .. seealso::
9
10 :ref:`ima-menu-operations` for more information on operations that can be performed
11 on images, or :ref:`ima-menu-processing` for information on processing features on
12 images.
13
14 .. figure:: /images/shots/i_analysis.png
15
16 Screenshot of the "Analysis" menu.
17
18 When the "Image Panel" is selected, the menus and toolbars are updated to
19 provide image-related actions.
20
21 The "Analysis" menu allows you to perform various computations on the current
22 image or group of images. It also allows you to compute statistics, to compute
23 the centroid, to detect peaks, to detect contours, and so on.
24
25 .. note::
26
27 In DataLab vocabulary, an "analysis" is a feature that computes a scalar
28 result from an image. This result is stored as metadata, and thus attached
29 to image. This is different from a "processing" which creates a new image
30 from an existing one.
31
32 Statistics
33 ^^^^^^^^^^
34
35 Compute statistics on selected image and show a summary table.
36
37 .. figure:: /images/shots/i_stats.png
38
39 Example of statistical summary table: each row is associated to an ROI
40 (the first row gives the statistics for the whole data).
41
42 Histogram
43 ^^^^^^^^^
44
45 Compute histogram of selected image and show it in the Signal Panel.
46
47 Parameters are:
48
49 .. list-table::
50 :header-rows: 1
51 :widths: 25, 75
52
53 * - Parameter
54 - Description
55 * - Bins
56 - Number of bins
57 * - Lower limit
58 - Lower limit of the histogram
59 * - Upper limit
60 - Upper limit of the histogram
61
62 .. figure:: /images/shots/i_histogram.png
63
64 Example of histogram.
65
66 Intensity profiles
67 ^^^^^^^^^^^^^^^^^^
68
69 Line profile
70 ~~~~~~~~~~~~
71
72 Extract an horizontal or vertical profile from each selected image, and create
73 new signals from these profiles.
74
75 .. figure:: /images/shots/i_profile.png
76
77 Line profile dialog. Parameters may also be set manually
78 ("Edit profile parameters" button).
79
80 Segment profile
81 ~~~~~~~~~~~~~~~
82
83 Extract a segment profile from each selected image, and create new signals
84 from these profiles.
85
86 Average profile
87 ~~~~~~~~~~~~~~~
88
89 Extract an horizontal or vertical profile averaged over a rectangular area, from
90 each selected image, and create new signals from these profiles.
91
92 .. figure:: /images/shots/i_profile_average.png
93
94 Average profile dialog: the area is defined by a rectangle shape.
95 Parameters may also be set manually ("Edit profile parameters" button).
96
97 Radial profile extraction
98 ~~~~~~~~~~~~~~~~~~~~~~~~~
99
100 Extract a radial profile from each selected image, and create new signals from
101 these profiles.
102
103 The following parameters are available:
104
105 .. list-table::
106 :header-rows: 1
107 :widths: 25, 75
108
109 * - Parameter
110 - Description
111 * - Center
112 - Center around which the radial profile is computed: centroid, image center, or user-defined
113 * - X
114 - X coordinate of the center (if user-defined), in pixels
115 * - Y
116 - Y coordinate of the center (if user-defined), in pixels
117
118 Horizontal and vertical projections
119 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
120
121 Compute horizontal and vertical projection profiles:
122
123 - Horizontal projection: Sum the pixel values along each row (projection on the x-axis).
124 - Vertical projection: Sum the pixel values along each column (projection on the y-axis).
125
126 Centroid
127 ^^^^^^^^
128
129 Compute the centroid of the image using a robust adaptive algorithm.
130
131 DataLab uses the :py:func:`sigima.tools.image.get_centroid_auto` function to estimate
132 the image centroid.
133
134 This function combines the robustness of a Fourier-based approach
135 (as discussed by `Weisshaar et al. <http://www.mnd-umwelttechnik.fh-wiesbaden.de/pig/weisshaar_u5.pdf>`_)
136 with fallback mechanisms designed to handle pathological cases, such as:
137
138 - truncated or asymmetric shapes,
139 - off-center objects,
140 - strong background noise.
141
142 Internally, :py:func:`sigima.tools.image.get_centroid_auto` compares the
143 Fourier result with two alternative estimators (a projected profile-based method
144 :py:func:`sigima.tools.image.get_projected_profile_centroid` and a standard centroid
145 from `scikit-image`), and selects the most consistent one.
146
147 This strategy ensures accurate and stable results across a wide range
148 of image types — from clean laboratory data to noisy or partial acquisitions.
149
150 Minimum enclosing circle center
151 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
152
153 Compute the circle contour enclosing image values above
154 a threshold level defined as the half-maximum value.
155
156 2D peak detection
157 ^^^^^^^^^^^^^^^^^
158
159 Automatically find peaks on image using a minimum-maximum filter algorithm.
160
161 .. figure:: /images/shots/i_peak2d_test.png
162
163 Example of 2D peak detection.
164
165 .. seealso::
166 See :ref:`ref-to-2d-peak-detection` for more details on algorithm and associated parameters.
167
168 Contour detection
169 ^^^^^^^^^^^^^^^^^
170
171 Automatically extract contours and fit them using a circle or an ellipse,
172 or directly represent them as a polygon.
173
174 .. figure:: /images/shots/i_contour_test.png
175
176 Example of contour detection.
177
178 .. seealso::
179 See :ref:`ref-to-contour-detection` for more details on algorithm and associated parameters.
180
181 .. note:: Computed scalar results are systematically stored as metadata.
182 Metadata is attached to image and serialized with it when exporting
183 current session in a HDF5 file.
184
185 Circle Hough transform
186 ^^^^^^^^^^^^^^^^^^^^^^
187
188 Detect circular shapes using circle Hough transform
189 (implementation based on `skimage.transform.hough_circle_peaks <https://scikit-image.org/docs/stable/api/skimage.transform.html?highlight=hough#skimage.transform.hough_circle_peaks>`_).
190
191 .. _ima-blob-detection:
192
193 Blob detection
194 ^^^^^^^^^^^^^^
195
196 Blob detection (DOG)
197 Detect blobs using Difference of Gaussian (DOG) method
198 (implementation based on `skimage.feature.blob_dog <https://scikit-image.org/docs/stable/api/skimage.feature.html#blob-dog>`_).
199
200 Blob detection (DOH)
201 Detect blobs using Determinant of Hessian (DOH) method
202 (implementation based on `skimage.feature.blob_doh <https://scikit-image.org/docs/stable/api/skimage.feature.html#blob-doh>`_).
203
204 Blob detection (LOG)
205 Detect blobs using Laplacian of Gaussian (LOG) method
206 (implementation based on `skimage.feature.blob_log <https://scikit-image.org/docs/stable/api/skimage.feature.html#blob-log>`_).
207
208 Blob detection (OpenCV)
209 Detect blobs using OpenCV implementation of `SimpleBlobDetector <https://docs.opencv.org/3.4/d0/d7a/classcv_1_1SimpleBlobDetector.html>`_.
210
211 .. note::
212 **Automatic ROI creation for detection features**
213
214 All detection features (2D peak detection, contour detection, circle Hough
215 transform, and blob detection) support automatic ROI creation around detected
216 objects. This feature:
217
218 - Creates rectangular or circular ROIs around each detected feature
219 - Automatically sizes ROIs based on minimum distance between detections
220 - Requires at least 2 detected objects to determine appropriate ROI size
221 - Enables subsequent processing on individual detected regions
222
223 To use this feature, enable "Create regions of interest" in the detection
224 parameters dialog and choose the desired ROI geometry.
225
226 Show results
227 ^^^^^^^^^^^^
228
229 Show the results of all analyses performed on the selected images. This shows the
230 same table as the one shown after having performed a computation.
231
232 Results label
233 ^^^^^^^^^^^^^
234
235 Toggle the visibility of result labels on the plot. When enabled, this checkable
236 menu item displays result annotations (such as centroid markers, detected contours,
237 blob circles, or other analysis shapes) directly on the image plot.
238
239 This option is synchronized between Signal and Image panels and persists across
240 sessions. It is only enabled when results are available for the selected image.
241
242 Plot results
243 ^^^^^^^^^^^^
244
245 Plot the results of analyses performed on the selected images, with user-defined
246 X and Y axes (e.g. plot the contour circle radius as a function of the image number).