Files

Powei Feng 541a4feae1 imgdiff: add positional and frequency robustness parameters (#9822 )

- Introduce `shiftRadius` to allow positional tolerances by searching
  a local neighborhood, absorbing sub-pixel shifts and MSAA quirks.
- Introduce `blurRadius` to apply local area averaging, ignoring
  high-frequency noise like hardware dithering.
- Enhance `ImageDiffResult` to include an `averageError` array and
  a 10-bin `errorHistogram` for actionable failure debugging.
- Update Android JNI bindings (`ImageDiff.java` and `ImageDiff.cpp`)
  to propagate the new error distribution statistics to Java callers.
- Update C++ unit tests to cover the new heuristic options.
- Document the new parameters and JSON result format in README.md.
- Add synthetic image generation tests in `tools/diffimg/tests/` to
  validate the CLI tool's handling of spatial shifts and dithering.

2026-03-25 00:31:18 +00:00

config_blur.json

imgdiff: add positional and frequency robustness parameters (#9822 )

2026-03-25 00:31:18 +00:00

config_shift.json

imgdiff: add positional and frequency robustness parameters (#9822 )

2026-03-25 00:31:18 +00:00

config_strict.json

imgdiff: add positional and frequency robustness parameters (#9822 )

2026-03-25 00:31:18 +00:00

gen_images.py

imgdiff: add positional and frequency robustness parameters (#9822 )

2026-03-25 00:31:18 +00:00

README.md

imgdiff: add positional and frequency robustness parameters (#9822 )

2026-03-25 00:31:18 +00:00

README.md

DiffImg Python Tests

This directory contains a suite of synthetic image tests to validate the diffimg tool's robustness configurations (shiftRadius and blurRadius).

Files

gen_images.py: A Python script that generates synthetic PPM images.
- ref.ppm: A reference image of a white circle.
- cand_shift.ppm: The same circle shifted by 1 pixel horizontally.
- cand_blur.ppm: The reference circle with high-frequency dithering noise added.
config_strict.json: An exact-match configuration (maxAbsDiff = 0.01).
config_shift.json: A configuration with a 1-pixel shift tolerance (shiftRadius = 1).
config_blur.json: A configuration with a local area average check (blurRadius = 1).

How to Run

Because diffimg depends on libs/imageio, which may use varying decoders based on OS capabilities, it is recommended to test with PNG files.

Prerequisites:

Python 3
sips (macOS native) or ImageMagick (for Linux/Windows) to convert PPM to PNG.
A compiled diffimg binary.

Steps (macOS Example):

Navigate to this directory:
```
cd tools/diffimg/tests/
```
Generate the PPM images:
```
python3 gen_images.py
```

Convert the generated PPM images to PNG (diffimg natively handles PNG cross-platform without needing LibTIFF or specific backends):

sips -s format png ref.ppm --out ref.png
sips -s format png cand_shift.ppm --out cand_shift.png
sips -s format png cand_blur.ppm --out cand_blur.png

Run the validation checks using the compiled binary (assuming it's built in out/cmake-release at the project root):

# Test spatial shift (should FAIL with strict, PASS with shift)
../../../out/cmake-release/tools/diffimg/diffimg -c config_strict.json ref.png cand_shift.png
../../../out/cmake-release/tools/diffimg/diffimg -c config_shift.json ref.png cand_shift.png

# Test high-frequency noise (should FAIL with strict, PASS with blur)
../../../out/cmake-release/tools/diffimg/diffimg -c config_strict.json ref.png cand_blur.png
../../../out/cmake-release/tools/diffimg/diffimg -c config_blur.json ref.png cand_blur.png