Back to search

University of the Aegean

Document image segmentation & text localization

Abstract

dc:description

In this work various issues related to the segmentation of pages and the extraction of textual information from document images are investigated. Novel approaches to challenging problems are presented, namely a layout analysis method, a text localization technique as well as a word-spotting and text retrieval system. Two software tools have been implemented in order to test the proposed algorithms. Results of detailed experiments are also shown. In Chapter 2, a method for the segmentation of complex (newspapers, magazines etc.) layouts is presented. No a priori knowledge of the page format is needed. Morphological operations are applied to both the foreground and the background, in order to connect neighboring regions and detect separator lines and columns respectively. Contour tracing is used for the extraction of shape information and classification of the connected components. In Chapter 3, a hybrid method for real-time localization of text embedded in images is proposed. It combines edge detection, morphological operators and a set of criteria based on spatial and geometrical features of the connected components. In Chapter 4, a technique appropriate for extracting all the textual information from documents with complex layouts is proposed. It integrates the foreground analysis part of the layout independent method presented in Chapter 2 with the fast and reliable method for text localization presented in Chapter 3. The first one is used to segment the page in text and image blocks while the second one is used to detect text that may be embedded inside the images. In Chapter 5, a classification-free word-spotting system, appropriate for the retrieval of printed historical document images is proposed. The system skips many of the procedures of a common approach. It does not include segmentation, feature extraction or classification. Instead it treats the queries as compact shapes and uses image processing techniques in order to localize a query in the document images. In Chapter 6, a novel technique for text retrieval is proposed. Although, it is inspired by the word spotting technique presented in Chapter 5, it spots characters, which makes the text retrieval more robust and permits the text queries. Chapter 7 summarizes the conclusions from all previous chapters, while in the appendixes three more applications of the proposed segmentation and text detection algorithms are explored: comic page segmentation, scene text localization and video text localization.

Degree

thesis:*
Grantor dc:publisher
University of the Aegean
Year dc:date
2016

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Vasilopoulos, Nikos
  • Βασιλόπουλος, Νίκος

Subjects

dc:subject × 8

Rights

dc:rights
Statement dc:rights
  • BY
Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/41035

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Vasilopoulos, Nikos; Βασιλόπουλος, Νίκος. Document image segmentation & text localization. University of the Aegean, 2016. http://hdl.handle.net/10442/hedi/41035