`pcv.roi.filter` plotting bugfix within `roi_type = "largest"` functionality

[HaleySchuhl]

Describe your changes Previously we were manually checking for hierarchy of contours and then we would draw the parent contour white, and carve out the child contour shape with black by plotting the first level hierarchy. This means that child of children contours are lost, plus we noticed a bug where the shape of the child contour is slightly enlarged by re-drawing it this way, regardless of settings to cv2.drawContours it still seemed like there was a line thickness that would change the result of cv2.countNonZero(filtered_mask) and we don't want to change the shape/size of masks by redrawing them. Instead, now we use the maxDepth=2 setting within drawing, provide the function all filtered contours, and just provide the contourIdx of the contour we ID as being the largest after ROI filtering. Removes the logic which had been replicating opencv contour and hierarchy datatype syntax.

Also edit to the unit test which is what brought my attention to this bugfix. Now assert that the pixel count of the largest shape is the same pre & post filtering, and add a secondary small object to get filtered out.

Type of update Is this a:

• Bug fix

Associated issues

• 1507 discovered although separate to the length vs area issue in logic.

• Additional context Add any other context about the problem here.

For the reviewer See this page for instructions on how to review the pull request.

• [x] PR functionality reviewed in a Jupyter Notebook
• [x] All tests pass
• [x] Test coverage remains 100%
• [x] Documentation tested
• [ ] New documentation pages added to plantcv/mkdocs.yml
• [ ] Changes to function input/output signatures added to updating.md
• [x] Code reviewed
• [x] PR approved

Test on simulated data to see the difference before & after

from plantcv.plantcv import Objects
import numpy as np
import cv2
from plantcv import plantcv as pcv

"""Test for PlantCV."""
# Create test data
img = np.zeros((100, 100), dtype=np.uint8)
mask = np.zeros(np.shape(img)[:2], dtype=np.uint8)
cnt = [np.array([[[25, 25]], [[25, 49]], [[49, 49]], [[49, 25]]], dtype=np.int32),
       np.array([[[34, 35]], [[35, 34]], [[39, 34]], [[40, 35]], [[40, 39]], [[39, 40]], [[35, 40]], [[34, 39]]],
                dtype=np.int32)]
cnt_str = np.array([[[-1, -1,  1, -1], [-1, -1, -1,  0]]], dtype=np.int32)
cv2.drawContours(mask, cnt, -1, (255), -1, lineType=8, hierarchy=cnt_str)
pcv.plot_image(mask)

# Add small square that should get filtered out
mask = cv2.rectangle(mask, (5,5), (7,7), 255, -1)
pcv.plot_image(mask)
print(cv2.countNonZero(mask))

area_total = cv2.countNonZero(mask)

pcv.params.debug = "plot"
roi_r = pcv.roi.rectangle(mask, x=0, y=0, h=99, w=99)
roi = [np.array([[[0, 0]], [[0, 99]], [[99, 99]], [[99, 0]]], dtype=np.int32)]
roi_str = np.array([[[-1, -1, -1, -1]]], dtype=np.int32)
roi_Obj = Objects(contours=[roi], hierarchy=[roi_str])
filtered_mask = pcv.roi.filter(mask=mask, roi=roi_Obj, roi_type="largest")
print(cv2.countNonZero(filtered_mask))

[deepsource-io[bot]]

Here's the code health analysis summary for commits 8e34132..559d57c. View details on DeepSource ↗.

Analysis Summary

Analyzer	Status	Summary	Link
Python	✅ Success		View Check ↗
Test coverage	✅ Success		View Check ↗

Code Coverage Report

Metric	Aggregate	Python
Branch Coverage	100%	100%
Composite Coverage	99.7%	99.7%
Line Coverage	99.7%	99.7%
New Branch Coverage	100%	100%
New Composite Coverage	100%	100%
New Line Coverage	100%, ✅ Above Threshold	100%, ✅ Above Threshold

---

💡 If you’re a repository administrator, you can configure the quality gates from the settings.

[nfahlgren]

Changed merged in previous PR