RUS  ENG
Full version
JOURNALS // Avtomatika i Telemekhanika // Archive

Avtomat. i Telemekh., 2021 Issue 11, Pages 3–15 (Mi at15825)

Topical issue (end)

Character skeleton as a pen trace model for recognition from reconstructed trace

S. P. Arseev, L. M. Mestetskiy

Lomonosov Moscow State University, Moscow, 119991 Russia

Abstract: The article studies the handwriting recognition problem and the reduction of the text recognition problem from a text image to the text recognition problem from the trace of a pen writing the text. We propose a method based on a medial representation and reconstructing the pen trace from the handwritten text image. The method is substantiated based on an experimental study of character recognition from character images and from the reconstructed and true pen traces.

Keywords: deep learning, convolutional neural network, recurrent neural network, skeleton, character recognition, handwriting recognition.

Presented by the member of Editorial Board: A. A. Lazarev

Received: 20.01.2021
Revised: 15.06.2021
Accepted: 30.06.2021

DOI: 10.31857/S0005231021110015


 English version:
Automation and Remote Control, 2021, 82:11, 1835–1845

Bibliographic databases:


© Steklov Math. Inst. of RAS, 2024