Mercurial > repos > imgteam > points2binaryimage
changeset 4:4c0f16fb9f8d draft default tip
planemo upload for repository https://github.com/BMCV/galaxy-image-analysis/tree/master/tools/points2binaryimage/ commit a3c77a79db469c9ad18b666f2c64d6e8f573945f
| author | imgteam |
|---|---|
| date | Wed, 25 Sep 2024 08:29:59 +0000 |
| parents | a80f92ec6f89 |
| children | |
| files | creators.xml points2binaryimage.py points2binaryimage.xml test-data/input1.tsv test-data/input2.tsv test-data/out.tiff test-data/output1.original.tif test-data/output1.tif test-data/output2.tif test-data/points.tsv tests.xml |
| diffstat | 11 files changed, 260 insertions(+), 58 deletions(-) [+] |
line wrap: on
line diff
--- a/creators.xml Wed Jun 26 08:35:56 2024 +0000 +++ b/creators.xml Wed Sep 25 08:29:59 2024 +0000 @@ -5,6 +5,11 @@ <yield /> </xml> + <xml name="creators/rmassei"> + <person givenName="Riccardo" familyName="Massei"/> + <yield/> + </xml> + <xml name="creators/alliecreason"> <person givenName="Allison" familyName="Creason"/> <yield/>
--- a/points2binaryimage.py Wed Jun 26 08:35:56 2024 +0000 +++ b/points2binaryimage.py Wed Sep 25 08:29:59 2024 +0000 @@ -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, invert_xy=False): +def find_column(df: pd.DataFrame, candidates: List[str]) -> str: + """ + Returns the 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: - if has_header: - df = pd.read_csv(point_file, skiprows=1, header=None, delimiter="\t") - 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])) + # Read the tabular file with information from the header + if has_header: + 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) - if invert_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 + # Read the tabular file without header + else: + 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? @@ -41,14 +80,14 @@ if __name__ == "__main__": parser = argparse.ArgumentParser() - parser.add_argument('point_file', type=argparse.FileType('r'), help='label file') + parser.add_argument('point_file', type=argparse.FileType('r'), help='point file') parser.add_argument('out_file', type=str, help='out file (TIFF)') parser.add_argument('shapex', type=int, help='shapex') parser.add_argument('shapey', type=int, help='shapey') - parser.add_argument('--has_header', dest='has_header', default=False, help='set True if CSV has header') - parser.add_argument('--invert_xy', dest='invert_xy', default=False, help='invert x and y in CSV') + parser.add_argument('--has_header', dest='has_header', default=False, help='set True if point file has header') + parser.add_argument('--swap_xy', dest='swap_xy', default=False, help='Swap X and Y coordinates') args = parser.parse_args() # TOOL - points2binaryimage(args.point_file.name, args.out_file, [args.shapey, args.shapex], has_header=args.has_header, invert_xy=args.invert_xy) + points2binaryimage(args.point_file.name, args.out_file, (args.shapey, args.shapex), has_header=args.has_header, swap_xy=args.swap_xy)
--- a/points2binaryimage.xml Wed Jun 26 08:35:56 2024 +0000 +++ b/points2binaryimage.xml Wed Sep 25 08:29:59 2024 +0000 @@ -2,8 +2,9 @@ <description></description> <macros> <import>creators.xml</import> - <token name="@TOOL_VERSION@">0.2</token> - <token name="@VERSION_SUFFIX@">3</token> + <import>tests.xml</import> + <token name="@TOOL_VERSION@">0.3</token> + <token name="@VERSION_SUFFIX@">0</token> </macros> <creator> <expand macro="creators/bmcv" /> @@ -15,45 +16,68 @@ <xref type="bio.tools">galaxy_image_analysis</xref> </xrefs> <requirements> - <requirement type="package" version="0.14.2">scikit-image</requirement> - <requirement type="package" version="1.15.4">numpy</requirement> - <requirement type="package" version="0.23.4">pandas</requirement> - <requirement type="package" version="2018.7">pytz</requirement><!--pandas seems to have additional requirements that are not automatically installed--> - <requirement type="package" version="2.5.2">python-dateutil</requirement> - <requirement type="package" version="0.15.1">tifffile</requirement> + <requirement type="package" version="0.21">scikit-image</requirement> + <requirement type="package" version="1.26.4">numpy</requirement> + <requirement type="package" version="1.2.4">pandas</requirement> + <requirement type="package" version="2024.6.18">tifffile</requirement> </requirements> - <command detect_errors="aggressive"> - <![CDATA[ - python '$__tool_directory__/points2binaryimage.py' '$input' '$output' $shapex $shapey $has_header $invert_xy - ]]> - </command> + <command detect_errors="aggressive"><![CDATA[ + + python '$__tool_directory__/points2binaryimage.py' + '$input' + '$output' + $shapex + $shapey + $has_header + $swap_xy + + ]]></command> <inputs> - <param name="input" type="data" format="tabular" label="CSV point file"/> - <param name="shapex" type="integer" value="500" optional="true" min="1" max="2000" label="Width of output image" /> - <param name="shapey" type="integer" value="500" optional="true" min="1" max="2000" label="Height of output image" /> - <param name="has_header" type="boolean" checked="false" truevalue="--has_header True" falsevalue="" optional="true" label="Does point file contain header?" /> - <param name="invert_xy" type="boolean" checked="false" falsevalue="" truevalue="--invert_xy True" optional="true" label="Inverts x and y in CSV point file" /> + <param name="input" type="data" format="tabular" label="Tabular list of points"/> + <param name="shapex" type="integer" value="500" min="1" label="Width of output image" /> + <param name="shapey" type="integer" value="500" min="1" label="Height of output image" /> + <param name="has_header" type="boolean" checked="false" truevalue="--has_header True" falsevalue="" optional="true" label="Tabular list of points has header" /> + <param name="swap_xy" type="boolean" checked="false" falsevalue="" truevalue="--swap_xy True" optional="true" label="Swap X and Y coordinates" /> </inputs> <outputs> <data name="output" format="tiff" /> </outputs> <tests> + <!-- TSV without header --> <test> - <param name="input" value="points.tsv" /> - <param name="shapex" value="20" /> - <param name="shapey" value="30" /> + <param name="input" value="input1.tsv" /> + <param name="shapex" value="30" /> + <param name="shapey" value="20" /> <param name="has_header" value="false" /> - <param name="invert_xy" value="true" /> - <output name="output" ftype="tiff" file="out.tiff" compare="sim_size"/> + <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> - This tool converts a CSV list of points to a binary image by rasterizing the point coordinates. + **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). - Pixels not corresponding to any points in the CSV are assigned the value 0 (black). + 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>
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/test-data/input1.tsv Wed Sep 25 08:29:59 2024 +0000 @@ -0,0 +1,9 @@ +11.7555970149 10.4048507463 +15 14 +19 2 +5 4 +5 5 +5 6 +5 7 +5 8 +5 9
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/test-data/input2.tsv Wed Sep 25 08:29:59 2024 +0000 @@ -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
--- a/test-data/points.tsv Wed Jun 26 08:35:56 2024 +0000 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,9 +0,0 @@ -11.7555970149 10.4048507463 -15 14 -19 2 -5 4 -5 5 -5 6 -5 7 -5 8 -5 9
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/tests.xml Wed Sep 25 08:29:59 2024 +0000 @@ -0,0 +1,95 @@ +<macros> + + <!-- Macros for verification of image outputs --> + + <xml + name="tests/binary_image_diff" + tokens="name,value,ftype,metric,eps" + token_metric="mae" + token_eps="0.01"> + + <output name="@NAME@" value="@VALUE@" ftype="@FTYPE@" compare="image_diff" metric="@METRIC@" eps="@EPS@" pin_labels="0"> + <assert_contents> + <has_image_n_labels n="2"/> + <yield/> + </assert_contents> + </output> + + </xml> + + <xml + name="tests/label_image_diff" + tokens="name,value,ftype,metric,eps,pin_labels" + token_metric="iou" + token_eps="0.01" + token_pin_labels="0"> + + <output name="@NAME@" value="@VALUE@" ftype="@FTYPE@" compare="image_diff" metric="@METRIC@" eps="@EPS@" pin_labels="@PIN_LABELS@"> + <assert_contents> + <yield/> + </assert_contents> + </output> + + </xml> + + <xml + name="tests/intensity_image_diff" + tokens="name,value,ftype,metric,eps" + token_metric="rms" + token_eps="0.01"> + + <output name="@NAME@" value="@VALUE@" ftype="@FTYPE@" compare="image_diff" metric="@METRIC@" eps="@EPS@"> + <assert_contents> + <yield/> + </assert_contents> + </output> + + </xml> + + <!-- Variants of the above for verification of collection elements --> + + <xml + name="tests/binary_image_diff/element" + tokens="name,value,ftype,metric,eps" + token_metric="mae" + token_eps="0.01"> + + <element name="@NAME@" value="@VALUE@" ftype="@FTYPE@" compare="image_diff" metric="@METRIC@" eps="@EPS@" pin_labels="0"> + <assert_contents> + <has_image_n_labels n="2"/> + <yield/> + </assert_contents> + </element> + + </xml> + + <xml + name="tests/label_image_diff/element" + tokens="name,value,ftype,metric,eps" + token_metric="iou" + token_eps="0.01" + token_pin_labels="0"> + + <element name="@NAME@" value="@VALUE@" ftype="@FTYPE@" compare="image_diff" metric="@METRIC@" eps="@EPS@" pin_labels="@PIN_LABELS@"> + <assert_contents> + <yield/> + </assert_contents> + </element> + + </xml> + + <xml + name="tests/intensity_image_diff/element" + tokens="name,value,ftype,metric,eps" + token_metric="rms" + token_eps="0.01"> + + <element name="@NAME@" value="@VALUE@" ftype="@FTYPE@" compare="image_diff" metric="@METRIC@" eps="@EPS@"> + <assert_contents> + <yield/> + </assert_contents> + </element> + + </xml> + +</macros>