Effects of Preprocessing in Person Identification Using Ear Features
Conference proceedings article
Authors/Editors
Strategic Research Themes
Publication Details
Author list: Patmanee J., Kanprachar S., Chamnongthai K.
Publisher: Taiwan Association for Aerosol Research
Publication year: 2021
Start page: 443
End page: 447
Number of pages: 5
ISBN: 9781665411974
ISSN: 1680-8584
eISSN: 2071-1409
Languages: English-Great Britain (EN-GB)
Abstract
Currently, face masking is strongly recommended for people when being outside in order to prevent the COVID-19 spread. However, by doing so, the face area is significantly blocked by the mask, resulting in an ineffective accuracy for face recognition system. To be able to identify a person while wearing a face mask, an alternative system has to be considered. There have been several studies in ear recognition system in which an impressive accuracy is obtained. In this work, ear recognition system with the AMI ear database is studied.The feature in terms of histogram of oriented gradients (HOG) is used, and the support vector machine (SVM) is adopted for classification process. To increase the recognition accuracy, ear images are preprocessed by adjusting the sharpness level. It is found that using the concatenated HOG features from the sharpened RGB and HSV images, a promising average recognition accuracy of 86% and the standard deviation of 2.91% are obtained. © 2021 IEEE.
Keywords
AMI Ear Database, Ear recognition system, Histogram of Oriented Gradients (HOG)