Add support for arbitrary columns and fix bugs

tools/points2binaryimage/points2binaryimage.py

@@ -1,36 +1,75 @@
 import argparse
 import os
 import warnings
+from typing import List
 
 import numpy as np
 import pandas as pd
+import scipy.ndimage as ndi
 import skimage.io
 
 
-def points2binaryimage(point_file, out_file, shape=[500, 500], has_header=False, swap_xy=False):
+def find_column(df: pd.DataFrame, candidates: List[str]) -> str:
+    """
+    Returns tje column name present in `df` and the list of `candidates`.
 
-    img = np.zeros(shape, dtype=np.int16)
+    Raises:
+        KeyError: If there is no candidate column name present in `df`, or more than one.
+    """
+    intersection = frozenset(df.columns) & frozenset(candidates)
+    if len(intersection) == 0:
+        raise KeyError(f'No such column: {", ".join(candidates)}')
+    elif len(intersection) > 1:
+        raise KeyError(f'The column names {", ".join(intersection)} are ambiguous')
+    else:
+        return next(iter(intersection))
+
+
+def points2binaryimage(point_file, out_file, shape, has_header=False, swap_xy=False, bg_value=0, fg_value=0xffff):
+
+    img = np.full(shape, dtype=np.uint16, fill_value=bg_value)
     if os.path.exists(point_file) and os.path.getsize(point_file) > 0:
+
+        # Read the tabular file with information from the header
         if has_header:
-            df = pd.read_csv(point_file, skiprows=1, header=None, delimiter="\t")
+            df = pd.read_csv(point_file, delimiter='\t')
+            pos_x_column = find_column(df, ['pos_x', 'POS_X'])
+            pos_y_column = find_column(df, ['pos_y', 'POS_Y'])
+            pos_x_list = df[pos_x_column].round().astype(int)
+            pos_y_list = df[pos_y_column].round().astype(int)
+            assert len(pos_x_list) == len(pos_y_list)
+            try:
+                radius_column = find_column(df, ['radius', 'RADIUS'])
+                radius_list = df[radius_column]
+            except KeyError:
+                radius_list = [0] * len(pos_x_list)
+
+        # Read the tabular file without header
         else:
-            df = pd.read_csv(point_file, header=None, delimiter="\t")
-
-        for i in range(0, len(df)):
-            a_row = df.iloc[i]
-            if int(a_row[0]) < 0 or int(a_row[1]) < 0:
-                raise IndexError("Point {},{} is out of image with bounds {},{}.".format(int(a_row[0]), int(a_row[1]), shape[0], shape[1]))
-
-            if swap_xy:
-                if img.shape[0] <= int(a_row[0]) or img.shape[1] <= int(a_row[1]):
-                    raise IndexError("Point {},{} is out of image with bounds {},{}.".format(int(a_row[0]), int(a_row[1]), shape[0], shape[1]))
-                else:
-                    img[int(a_row[1]), int(a_row[0])] = 32767
+            df = pd.read_csv(point_file, header=None, delimiter='\t')
+            pos_x_list = df[0].round().astype(int)
+            pos_y_list = df[1].round().astype(int)
+            assert len(pos_x_list) == len(pos_y_list)
+            radius_list = [0] * len(pos_x_list)
+
+        # Optionally swap the coordinates
+        if swap_xy:
+            pos_x_list, pos_y_list = pos_y_list, pos_x_list
+
+        # Perform the rasterization
+        for y, x, radius in zip(pos_y_list, pos_x_list, radius_list):
+
+            if y < 0 or x < 0 or y >= shape[0] or x >= shape[1]:
+                raise IndexError(f'The point x={x}, y={y} exceeds the bounds of the image (width: {shape[1]}, height: {shape[0]})')
+
+            if radius > 0:
+                mask = np.ones(shape, dtype=bool)
+                mask[y, x] = False
+                mask = (ndi.distance_transform_edt(mask) <= radius)
+                img[mask] = fg_value
             else:
-                if img.shape[0] <= int(a_row[1]) or img.shape[1] <= int(a_row[0]):
-                    raise IndexError("Point {},{} is out of image with bounds {},{}.".format(int(a_row[1]), int(a_row[0]), shape[0], shape[1]))
-                else:
-                    img[int(a_row[0]), int(a_row[1])] = 32767
+                img[y, x] = fg_value
+
     else:
         raise Exception("{} is empty or does not exist.".format(point_file))  # appropriate built-in error?
 
@@ -51,4 +90,4 @@ def points2binaryimage(point_file, out_file, shape=[500, 500], has_header=False,
     args = parser.parse_args()
 
     # TOOL
-    points2binaryimage(args.point_file.name, args.out_file, [args.shapey, args.shapex], has_header=args.has_header, swap_xy=args.swap_xy)
+    points2binaryimage(args.point_file.name, args.out_file, (args.shapey, args.shapex), has_header=args.has_header, swap_xy=args.swap_xy)

tools/points2binaryimage/points2binaryimage.xml

@@ -3,8 +3,8 @@
     <macros>
         <import>creators.xml</import>
         <import>tests.xml</import>
-        <token name="@TOOL_VERSION@">0.2</token>
-        <token name="@VERSION_SUFFIX@">3</token>
+        <token name="@TOOL_VERSION@">0.3</token>
+        <token name="@VERSION_SUFFIX@">0</token>
     </macros>
     <creator>
         <expand macro="creators/bmcv" />
@@ -46,22 +46,39 @@
         <!-- TSV without header -->
         <test>
             <param name="input" value="input1.tsv" />
-            <param name="shapex" value="20" /> 
-            <param name="shapey" value="30" />
+            <param name="shapex" value="30" />
+            <param name="shapey" value="20" />
             <param name="has_header" value="false" />
             <param name="swap_xy" value="true" />
             <expand macro="tests/binary_image_diff" name="output" value="output1.tif" ftype="tiff" />
         </test>
         <!-- TSV with header -->
+        <test>
+            <param name="input" value="input2.tsv" />
+            <param name="shapex" value="205" />
+            <param name="shapey" value="84" />
+            <param name="has_header" value="true" />
+            <param name="swap_xy" value="false" />
+            <expand macro="tests/binary_image_diff" name="output" value="output2.tif" ftype="tiff" />
+        </test>
     </tests>
     <help>
 
         **Converts a tabular list of points to a binary image by rasterizing the point coordinates.**
 
         The created image is a single-channel image with 16 bits per pixel (unsigned integer).
-        The points are rasterized with value 32767 (white).
+        The points are rasterized with value 65535 (white).
         Pixels not corresponding to any points in the tabular file are assigned the value 0 (black).
 
+        The tabular list of points can either be header-less.
+        In this case, the first and second columns are expected to be the X and Y coordinates, respectively.
+        Otherwise, if a header is present, it is searched for the following column names:
+
+        - ``pos_x`` or ``POS_X``: This column corresponds to the X coordinates.
+        - ``pos_y`` or ``POS_Y``: This column corresponds to the Y coordinates.
+        - If a ``radius`` or ``RADIUS`` column is present,
+          then the points will be rasterized as circles of the corresponding radii.
+
     </help>
     <citations>
         <citation type="doi">10.1016/j.jbiotec.2017.07.019</citation> 

tools/points2binaryimage/test-data/input2.tsv

@@ -0,0 +1,39 @@
+frame	pos_x	pos_y	scale	radius	intensity
+1	85	32	1.33	3.77	18807.73
+1	190	25	1.78	5.03	24581.44
+1	137	26	1.44	4.09	19037.59
+1	63	42	1.44	4.09	22390.80
+1	107	44	1.33	3.77	23429.96
+1	61	27	1.56	4.40	18052.18
+1	158	39	1.44	4.09	18377.02
+1	190	14	1.33	3.77	18548.86
+1	182	33	1.78	5.03	26467.79
+1	39	39	1.44	4.09	14782.43
+1	169	26	1.33	3.77	14203.41
+1	61	54	1.33	3.77	23248.06
+1	95	52	1.33	3.77	21480.71
+1	23	60	1.89	5.34	25203.43
+1	84	24	1.56	4.40	16630.57
+1	121	47	1.67	4.71	15459.11
+1	66	49	1.11	3.14	23858.07
+1	115	36	2.00	5.66	16389.10
+1	55	51	1.33	3.77	23548.90
+1	130	72	1.67	4.71	15769.02
+1	117	23	1.33	3.77	16763.14
+1	45	52	1.56	4.40	22877.61
+1	36	71	1.56	4.40	20780.96
+1	78	17	1.33	3.77	16844.51
+1	101	38	1.56	4.40	21376.59
+1	147	31	1.78	5.03	16597.14
+1	163	55	2.00	5.66	18301.54
+1	164	23	1.33	3.77	17073.82
+1	150	24	1.56	4.40	15440.02
+1	151	67	1.78	5.03	18419.96
+1	26	53	2.00	5.66	20586.01
+1	79	62	1.33	3.77	15232.88
+1	69	17	1.11	3.14	15601.83
+1	83	52	1.33	3.77	18315.00
+1	16	54	2.00	5.66	22140.66
+1	166	61	1.78	5.03	18488.78
+1	163	43	1.44	4.09	16925.49
+1	130	53	1.78	5.03	15101.96