2021/11/17 by Hélder Campos, Campos, Hélder, Nuno Paulino +1
Computer Science · #Handwritten Text Recognition Techniques #Text and Document Classification Technologies #Natural Language Processing Techniques
paper · pdf · doi:10.48550/arxiv.2111.10204
This paper presents results of a study of the performance of several base\nclassifiers for recognition of handwritten characters of the modern Latin\nalphabet. Base classification performance is further enhanced by utilizing\nViterbi error correction by determining the Viterbi sequence. Hidden Markov\nModels (HMMs) models exploit relationships between letters within a word to\ndetermine the most likely sequence of characters. Four base classifiers are\nstudied along with eight feature sets extracted from the handwritten dataset.\nThe best classification performance after correction was 89.8%, and the average\nwas 68.1%\n