Please use this identifier to cite or link to this item:
Title: Handwritten Document Image Retrieval
Authors: ZHANG XI
Keywords: Text line segmentation, handwritten word recognition, heat kernel signature, keyword spotting, handwritten document image retrieval
Issue Date: 14-Aug-2014
Citation: ZHANG XI (2014-08-14). Handwritten Document Image Retrieval. ScholarBank@NUS Repository.
Abstract: Information retrieval for handwritten document images is more challenging due to the difficulties in complex layout analysis, large variations of writing styles, and degradation or low quality of historical manuscripts. In order to overcome the limitations of OCR, keyword spotting becomes an alternative way to retrieve handwritten documents. It only needs the features extracted from the imaged documents. In view of large variations in handwriting styles, this thesis will first present a method for extracting text lines from handwritten documents. Secondly, a combination of two well-trained networks is used to improve the recognition performance for word images. Thirdly, Heat Kernel Signature (HKS) is used to represent the documents, and to achieve word image matching and segmentation-free keyword spotting. Finally, we will present our method for writer identification and content relevance retrieval.
Appears in Collections:Ph.D Theses (Open)

Show full item record
Files in This Item:
File Description SizeFormatAccess SettingsVersion 
ThesisFinalSubmission.pdf6.17 MBAdobe PDF



Page view(s)

checked on Oct 14, 2021


checked on Oct 14, 2021

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.