Complementary features combined in an HMM-based system to recognize handwritten digits

Abstract

We combine complementary features based on foreground and background information in an HMM-based classifier to recognize handwritten digits. A zoning scheme based on column and row models provides a way of dividing the digit into zones without making the features size variant. This strategy allows us to avoid the digit normalization, while it provides a way of having information from specific zones of the digit. Recognition rates around 98% have been achieved using 60,000 digit samples of the NIST SD19 database.