I built a small Python Reed-Solomon codec aimed at OCR workflows where printed codes may be degraded by dot-matrix printers, missing pins, fading ribbons, low resolution, dirt, or partial character damage.
It uses GF(256) Reed-Solomon correction plus an OCR-safe parity representation using a reduced alphabet that avoids confusing characters like 0/O, 1/I/l, 5/S, 8/B, etc. Each parity byte is encoded as two OCR-safe characters, so printed labels or IDs can be scanned, checked, and corrected when OCR makes a limited number of symbol mistakes.
The repo includes a ReedSolomonForOcr class, OCR-safe parity helpers, a demo, and unit tests with deterministic correction scenarios.
Dwedit•1h ago
Do we have any screenshots of what the text looks like, and what the error correction looks like?
chasangchual•3h ago
It uses GF(256) Reed-Solomon correction plus an OCR-safe parity representation using a reduced alphabet that avoids confusing characters like 0/O, 1/I/l, 5/S, 8/B, etc. Each parity byte is encoded as two OCR-safe characters, so printed labels or IDs can be scanned, checked, and corrected when OCR makes a limited number of symbol mistakes.
The repo includes a ReedSolomonForOcr class, OCR-safe parity helpers, a demo, and unit tests with deterministic correction scenarios.