Tesseract OCR Integration #10
Description
Description
Integrate the Tesseract OCR engine into the daemon, consuming preprocessed frames and outputting detected text strings. Includes confidence thresholding and handling of partial/noisy results.
Definition of Done
[ ] Tesseract C API (TessBaseAPI) initialized with English language data
[ ] Accepts a preprocessed image buffer and returns extracted text as a string
[ ] Confidence score is retrieved per-word; words below a configurable threshold are discarded
[ ] Handles edge cases: blank page (no text detected), very small text, mixed fonts
[ ] Tested against at least 10 sample images with ground truth; accuracy results documented
[ ] Average OCR processing time per frame documented (target: under 3 seconds on RPi4)
[ ] Source committed as ocr.c / ocr.h in the app repo
[ ] Memory is properly freed after each OCR pass (no leaks confirmed via manual review or valgrind)