Numeral Recognition by Weighting Local Decisions

This paper presents a new technique to improve the combination of classification decisions obtained from local analysis of patterns. Specifically, a genetic algorithm is used to determine the optimal weight vector to balance the local decisions in the combination process. The experimental results, carried out in the field of hand-written numeral recognition, demonstrate the effectiveness of the new technique.

• Type: Conference
• Year: 2003
• Where: ICDAR
• Authors: Giovanni Dimauro, Sebastiano Impedovo, Raffaele Modugno, Giuseppe Pirlo
• Keywords: Classification Decisions, Document Analysis, Hand-written Numeral Recognition, ICDAR 2003, Optimal Weight Vector