Aggaluk Jaoukaew. Personal Authentication for Mobile Device Based on Features of Backside Thumb. Master's Degree(Electronics and Telecommunication Engineering). King Mongkut's University of Technology Thonburi. KMUTT Library.. : King Mongkut's University of Technology Thonburi, 2008.
Personal Authentication for Mobile Device Based on Features of Backside Thumb
Abstract:
Mobile device is required to be highly secure because it contains the personal data of
the owner while it is easy to lose. Therefore, it should be accessed only by the owner or
an authorized person. This thesis proposes to use the features of the backside left thumb
for identifying owner or authorized person of a mobile phone. This method starts from
capturing the back left thumb by a mobile phone camera. After that, the thumb image is
automatically extracted based on a skin color detection technique. Then, the region of
interest (ROI) is automatically cropped based on the width and a specific ratio of the
nail's length and the thumb's width. The width and the top of the thumb are easily
found by differentiating the mean intensity along the columns, and differentiating the
column along the center of the thumb then resize the picture to the standard size to
increase the accuracy of identification. A code of the cropped ROI is calculated based
on the method proposed by Kumar et al., This includes the nonnalization, the
multichannel filtering based on the Real Gabor Function (RGF), and the generation of
code referred to here as "ThumbCode". Finally, the ThumbCode under consideration is
compared with the ThumbCode of an authorized person based on the errors between the
deformed ThumbCodes. The method was verified with 8 persons' thumb images taken
from 3 approximate ranges and experiments with special conditions including long nails,
painted nails and outdoor were carried out. The result revealed that under a normal light
condition the overall accuracy is 91.81 percent, while having 8.75 percent faulty accepts
and 8.11 percent faulty rejections.