An Efficient Method for Hand-Torn Document Reconstruction


Salih, Nbhan D. and Mohd Isa, Wan Noorshahida and Abid, Abdelouahab (2019) An Efficient Method for Hand-Torn Document Reconstruction. International Journal of Engineering Research and Technology, 12 (7). pp. 986-992. ISSN 0974-3154

[img] Text
41j.pdf - Published Version
Restricted to Repository staff only

Download (864kB)


In this paper, an efficient method for hand-torn document reconstruction is proposed. The proposed method comprises three major stages, which are preprocessing, feature extraction, and matching respectively. The preprocessing stage starts with extracting the green-channel of the input fragment, and morphological opening and binarization is performed, followed by contour extraction and simplification. A number of features, such as vertex angle, line distance and orientation are extracted in the feature extraction stage. Afterwards, the method searches for the best matching between two fragments, correct the rotation angle, and join those fragments, and this procedure is performed iteratively until the whole document is reconstructed. A simple but effective procedure is proposed to measure the performance of the proposed reconstruction method. Two criteria, namely True Positive Rate (TPR) and False Negative Rate (FNR) have been used to evaluate the performance of the proposed method. The experimental results show that the proposed algorithm achieves 0.9777 and 0.0223 average values in terms of TPR and FNR respectively.

Item Type: Article
Uncontrolled Keywords: Image Reconstruction; Torn Document; Feature Matching; Polygonal Approximation.
Subjects: Q Science > QA Mathematics > QA150-272.5 Algebra
Divisions: Faculty of Computing and Informatics (FCI)
Depositing User: Ms Suzilawati Abu Samah
Date Deposited: 27 Apr 2022 00:45
Last Modified: 27 Apr 2022 00:45


Downloads per month over past year

View ItemEdit (login required)