Please use this identifier to cite or link to this item: http://hdl.handle.net/10553/42042
Title: Two-steps perceptual important points estimator in 8-connected curves from handwritten signature
Authors: Ferrer, Miguel A. 
Diaz, Moises 
Carmona-Duarte, Cristina 
Keywords: 1203 Ciencia de los ordenadores
metadata.dc.subject.other: Handwritting segmentation
High curvature points
Curvatures estimation
Perceptual important points estimation
Issue Date: 2018
Abstract: Estimating the salient points in 8-connected curves from handwritten signature is a difficult task due to their relation to the writer neuromotor system. This paper faces up this topic proposing a two-steps perceptual important points estimation method: the first step estimates the sharper salient points by a curvature analysis at multiple scales, whereas the second step estimates the smoother salient points relying on circular shapes between estimated salient points in step one. In this approach, both sharper and smoother salient points represent the set of perceptual important points in an eight connected signature trajectory. Our validations, conducted on 2112 signatures from 132 users of the BiosecurID database, are focused on i) evaluating the number of estimated perceptual important points; ii) evaluating their locations in the trajectory and iii) evaluating the accuracy of the estimated duration of the signatures from the number of perceptual important points. The obtained results are encouraging for new developments in handwriting analysis based on this procedure.
URI: http://hdl.handle.net/10553/42042
DOI: 10.1109/IPTA.2017.8310077
Source: Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA), Montreal, QC, 2017, pp. 1-5
Appears in Collections:Artículos

Files in This Item:
File SizeFormat 
08310077.pdf593,78 kBAdobe PDFView/Open
Show full item record

SCOPUSTM   
Citations

1
checked on Jul 13, 2019

Page view(s)

3
checked on Jul 14, 2019

Download(s)

4
checked on Jul 14, 2019

Google ScholarTM

Check

Altmetric


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