1 Commits

Author SHA1 Message Date
z060142
2ac63718a9 Implement perceptual image hash deduplication for bubble processing
- Added `simple_bubble_dedup.py` module using perceptual hashing (pHash) to detect duplicate chat bubbles based on visual similarity.
- System keeps recent N (default 5) hashes and skips bubbles with Hamming distance below threshold (default 5).
- Integrated into `run_ui_monitoring_loop()`:
  - Hash is computed upon bubble snapshot capture.
  - Duplicate check occurs before message enqueue.
  - Sender info is optionally attached to matching hash entries after successful processing.
- Deduplication state is persisted in `simple_bubble_dedup.json`.
- F7 (`clear_history`) and F8 (`reset_state`) now also clear image-based hash history.
- Removed or commented out legacy `recent_texts` text-based deduplication logic.

This visual deduplication system reduces false negatives caused by slight text variations and ensures higher confidence in skipping repeated bubble interactions.
2025-05-16 02:02:31 +08:00