Type-hint coverage audit: ~75% overall, parameter annotations lagging in several modules

Bắt đầu với src/machinevisiontoolbox/cvfuncs.py, ImageCore.py, Sources.py, ImageRegionFeatures.py, BagOfWords.py và ImageBlobs.py, so sánh các annotation của chúng với quy ước typing hiện có của dự án và các module được annotation đầy đủ như mvtb_types.py. Thêm các annotation còn thiếu cho tham số bằng các dạng hiện đại được nêu, đồng thời giữ nguyên hành vi; hoàn tất khi đã xử lý sáu module có độ bao phủ thấp nhất.

Found 2026-08-10 while reviewing unmerged branches during a cleanup pass. An old typing branch (base commit 2022-10-18, predates the src/-layout restructure) attempted a broader typing pass but is no longer mergeable — its module paths no longer exist. Later, separate efforts did land on main (e.g. acc8222b, 524c7b8c, PR #32/fix/annotations-and-deprecation-warnings), so coverage today is real but partial, not complete.

An AST-based audit of parameter/return annotations across src/machinevisiontoolbox gives:

OVERALL: 54.3% params, 95.4% returns, 74.9% average

Return-type coverage is already strong almost everywhere (many modules at 100%); the gap is almost entirely on parameter annotations. Full per-module breakdown, sorted by average:

Module                       Param %   Return %   Avg %
cvfuncs.py                    100.0%      0.0%     50.0%
ImageCore.py                   30.9%     94.7%     62.8%
Sources.py                     38.0%     91.1%     64.6%
ImageRegionFeatures.py         31.0%    100.0%     65.5%
BagOfWords.py                  39.1%     95.7%     67.4%
ImageBlobs.py                  38.0%    100.0%     69.0%
ImageConstants.py              72.3%     71.4%     71.9%
ImageMultiview.py              68.2%     83.3%     75.8%
ImageIO.py                     58.5%     93.3%     75.9%
ImageWholeFeatures.py          55.6%     97.6%     76.6%
ImageProcessing.py             68.0%     86.4%     77.2%
ImagePointFeatures.py          55.6%    100.0%     77.8%
BundleAdjust.py                56.0%    100.0%     78.0%
VisualServo.py                 63.2%     93.2%     78.2%
Kernel.py                      57.5%    100.0%     78.8%
PointCloud.py                  57.9%    100.0%     78.9%
Camera.py                      61.3%     99.0%     80.1%
ImageColor.py                  60.9%    100.0%     80.4%
ImageFiducials.py              69.6%     94.1%     81.9%
ImageLineFeatures.py           66.7%    100.0%     83.3%
ImageReshape.py                72.8%     96.0%     84.4%
ImageMorph.py                  77.1%    100.0%     88.5%
ImageTensor.py                 77.8%    100.0%     88.9%
ImageSpatial.py                78.7%    100.0%     89.4%
camera_derivatives.py         100.0%    100.0%    100.0%
decorators.py                 100.0%    100.0%    100.0%
docbugs.py                    100.0%    100.0%    100.0%
fiducial.py                   100.0%    100.0%    100.0%
mvtb_types.py                 100.0%    100.0%    100.0%

Fix

Work through the lowest-param-coverage modules first (cvfuncs.py, ImageCore.py, Sources.py, ImageRegionFeatures.py, BagOfWords.py, ImageBlobs.py — all under 40% param coverage), adding modern X | Y / X | None / list[X] annotations per the project's typing convention. No architectural changes needed, just incremental annotation additions per module.