Prototype matching is a fundamental technique in pattern classification, where an unknown pattern is compared against a set of stored prototype patterns, and is assigned to the class of the most similar prototype.
- Each prototype represents a pattern class.
- Similarity is measured using various distance metrics.
Real-Life Application: Bank Cheque Recognition using E-13B Font
- In 1950s, banks used magnetic ink and stylized characters (E-13B font) for fast cheque processing.
- Each character’s waveform is sampled at 9 points → Forms a 9-D pattern vector.
- Classification is done using minimum-distance matching with pre-stored 9-D prototype vectors.
This enabled fast recognition using analog circuits without needing a full computer.