Answers are grounded in this paper and cite their sources.
A feature set for the encoding of optical images of handwritten characters is presented. It allows good recognition rates when coupled with a neural network classifier and can be easily implemented in analogue CMOS VLSI technology. A circuit for the extraction of such features is described.
No takes yet. Share an insight, caveat, or question.
Caviglia et al. (1994) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: