REAL-TIME RECOGNITION OF HANDPRINTED TEXT
Abstract
The report describes a scheme which permits an on-line computer user to print text naturally, and have it recognized accurately. This scheme, which recognizes 53 symbols, enables a user to communicate data and directives to a computer by using only a RAND tablet with its pen as an input device. The tablet provides high resolution point-by-point pen location data which the recognition scheme analyzes as a symbol is being drawn. This analysis describes a handprinted symbol primarily as a time sequence of features. Identification is based on a data-dependent sequence of tests. Symbols are separated from one another according to their identifications and relative positions. The scheme is evaluated by determining how accurately it recognizes isolated symbols, and how much time users need to compose computer code on-line.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 1966
- Accession Number
- AD0641252
Entities
People
- Gabriel F. Groner
Organizations
- RAND Corporation