dev(export):expose some data which is needed for exporting to EcoTaxa

Author: grololo06

src/ZooProcess_lib/Extractor.py

@@ -42,23 +42,29 @@ def extract_all_with_border_to_dir(
         for index, a_roi in enumerate(rois, 1):
             img = self.extract_image_at_ROI(image, a_roi, True)
             resized_img = self._add_border_and_legend(img, longline)
-            for extension in [".png"]:
-                img_filename = (
-                    naming_prefix + "_" + unique_visible_key(a_roi) + extension
-                )
-                img_path = destination_dir / img_filename
-                if extension == ".jpg":
-                    # Convert the image to 3 channels before saving, like legacy
-                    img_to_save = cv2.merge([resized_img, resized_img, resized_img])
-                else:
-                    # Save 1 byte/pixel, grey
-                    img_to_save = resized_img
+            img_filename = self.vignette_name(a_roi, naming_prefix)
+            img_path = destination_dir / img_filename
+            # Save 1 byte/pixel, grey
+            img_to_save = resized_img
+            save_jpg_or_png_image(
+                img_to_save,
+                resolution,
+                img_path,
+            )
+            if False:
+                # Convert the image to 3 channels before saving, like legacy
+                # Note: unreachable code kept for later
+                img_to_save_3chan = cv2.merge([resized_img, resized_img, resized_img])
                 save_jpg_or_png_image(
-                    img_to_save,
+                    img_to_save_3chan,
                     resolution,
-                    img_path,
+                    img_path[:-4] + ".jpg",
                 )
 
+    @staticmethod
+    def vignette_name(a_roi, naming_prefix):
+        return naming_prefix + "_" + unique_visible_key(a_roi) + ".png"
+
     def extract_all_to_images(
         self,
         image: np.ndarray,

src/ZooProcess_lib/Features.py

@@ -449,3 +449,41 @@ def legacy_measures_list_from_roi_list(
             Features(image, resolution, p, self.threshold).as_measures(only)
             for p in roi_list
         ]
+
+    def ecotaxa_measures_list_from_roi_list(
+        self,
+        image: ndarray,
+        resolution: int,
+        roi_list: List[ROI],
+    ) -> list[dict[str, int | float]]:
+        return [
+            Features(image, resolution, p, self.threshold).as_ecotaxa()
+            for p in roi_list
+        ]
+
+
+ECOTAXA_TSV_ROUNDINGS = {  # from Zooscan_print_pid_5.txt
+    "object_major": 1,
+    "object_minor": 1,
+    "object_angle": 1,
+    "object_feret": 1,
+    "object_x": 2,
+    "object_y": 2,
+    "object_xm": 2,
+    "object_ym": 2,
+    "object_perim.": 2,
+    "object_%area": 2,
+    "object_mean": 2,
+    "object_stddev": 3,
+    "object_circ.": 3,
+    "object_skew": 3,
+    "object_kurt": 3,
+    "object_fractal": 3,
+    "object_slope": 3,
+    "object_fcons": 3,
+    "object_symetrieh": 3,
+    "object_symetriev": 3,
+    "object_thickr": 3,
+}
+ECOTAXA_TSV_ROUNDINGS.update
+{"object_meanpos": 3}

src/ZooProcess_lib/Segmenter.py

@@ -60,16 +60,20 @@ def get_mask_from_image(self, image: ndarray) -> np.ndarray:
         # mask is black (255) objects on white (0) background, it's for human eye
         return mask
 
+    @staticmethod
+    def pixel_size(resolution):
+        return 25.4 / resolution
+
     def find_ROIs_in_image(
         self, image: ndarray, resolution: int, method: int = METH_TOP_CONTOUR_SPLIT
     ) -> Tuple[List[ROI], List[int]]:
         """:return: (list of ROIs, elimination statistics i.e. number of particles eliminated at different stages)"""
-        pixel = 25.4 / resolution
+        pixel_sz = self.pixel_size(resolution)
         sm_min = (3.1416 / 4) * pow(self.minsize, 2)
         sm_max = (3.1416 / 4) * pow(self.maxsize, 2)
         # part_size_* are in pixel^2
-        part_size_min: int = round(sm_min / (pow(pixel, 2)))
-        part_size_max: int = round(sm_max / (pow(pixel, 2)))
+        part_size_min: int = round(sm_min / (pow(pixel_sz, 2)))
+        part_size_max: int = round(sm_max / (pow(pixel_sz, 2)))
         assert image.dtype == np.uint8
         height, width = image.shape[:2]
         # Threshold the source image to have a b&w mask

tests/test_calculators.py

@@ -1,10 +1,9 @@
 from math import exp
 
-import cv2
 import numpy as np
 import pytest
 
-from ZooProcess_lib.Features import Features, TYPE_BY_ECOTAXA
+from ZooProcess_lib.Features import Features, TYPE_BY_ECOTAXA, ECOTAXA_TSV_ROUNDINGS
 from ZooProcess_lib.ROI import ecotaxa_tsv_unq
 from ZooProcess_lib.Segmenter import Segmenter
 from ZooProcess_lib.calculators.Custom import fractal_mp, ij_perimeter
@@ -68,33 +67,6 @@ def test_ij_like_perimeter():
     assert round(perim, 3) == 359.772
 
 
-ECOTAXA_TSV_ROUNDINGS = {  # from Zooscan_print_pid_5.txt
-    "object_major": 1,
-    "object_minor": 1,
-    "object_angle": 1,
-    "object_feret": 1,
-    "object_x": 2,
-    "object_y": 2,
-    "object_xm": 2,
-    "object_ym": 2,
-    "object_perim.": 2,
-    "object_%area": 2,
-    "object_mean": 2,
-    "object_stddev": 3,
-    "object_circ.": 3,
-    "object_skew": 3,
-    "object_kurt": 3,
-    "object_fractal": 3,
-    "object_slope": 3,
-    "object_fcons": 3,
-    "object_symetrieh": 3,
-    "object_symetriev": 3,
-    "object_thickr": 3,
-}
-ECOTAXA_TSV_ROUNDINGS.update
-{"object_meanpos": 3}
-
-
 @pytest.mark.parametrize(
     "img",
     [
@@ -106,7 +78,9 @@ def test_ij_like_features(img: str):
     image = loadimage(MEASURES_DIR / f"crop_{img}.png")
     THRESHOLD = 243
     segmenter = Segmenter(0.3, 100, THRESHOLD)
-    rois, _ = segmenter.find_ROIs_in_cropped_image(image, 2400, Segmenter.METH_TOP_CONTOUR_SPLIT)
+    rois, _ = segmenter.find_ROIs_in_cropped_image(
+        image, 2400, Segmenter.METH_TOP_CONTOUR_SPLIT
+    )
     assert len(rois) == 1
     feat = Features(image=image, resolution=2400, roi=rois[0], threshold=THRESHOLD)
     exp = read_ecotaxa_tsv(FEATURES_DIR / f"ecotaxa_{img}.tsv", TYPE_BY_ECOTAXA)