Detección de signos de puntuación en documentos de texto manuscrito

[EN] Handwriting recognition is a complex task and not always we obtain good results, either by the text conditions or by the kind of writing. That is why it remains research topic, where every time new techniques and methods are obtained . Currently text segmentation is done by line, as the recogni...

Descripción completa

Detalles Bibliográficos
Autor: Hernández Salmerón, Miguel
Tipo de recurso: tesis de maestría
Fecha de publicación:2017
País:España
Institución:Universitat Politècnica de València (UPV)
Repositorio:RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Idioma:español
OAI Identifier:oai:riunet.upv.es:10251/89809
Acceso en línea:https://riunet.upv.es/handle/10251/89809
Access Level:acceso abierto
Palabra clave:Técnicas de clasificación
Procesado de imagen
Procesado de documentos
Reconocimiento de texto manuscrito
LENGUAJES Y SISTEMAS INFORMATICOS
Máster Universitario en Inteligencia Artificial, Reconocimiento de Formas e Imagen Digital-Màster Universitari en Intel·ligència Artificial, Reconeixement de Formes i Imatge Digital
Descripción
Sumario:[EN] Handwriting recognition is a complex task and not always we obtain good results, either by the text conditions or by the kind of writing. That is why it remains research topic, where every time new techniques and methods are obtained . Currently text segmentation is done by line, as the recognition systems extract each line of the text and then analyze it. However, the extracted lines do not have to be coherent, since is normally does not coincide with a complete sentence. The goal of this project is to locate the punctuation marks found in the text to fragment it into complete sentences that have a coherent meaning. For this purpose, it has been decided to use a technique used to classify different types of images by using a convolutional neural network. The neural network has been trained with images of the different punctuation marks in order to recognize them throughout the text and to store the position in which they are. To make this project a manuscript of the year 1853 was used, from which the images of both the points and the commas have been obtained to generate the different corpus that have been used.