Share Email Print

Proceedings Paper

Robust non-parametric probabilistic image processing for face recognition and pattern recognition
Author(s): Meropi Pavlidou; George Zioutas
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

Face Recognition has been a pattern recognition application of great interest. Many mathematical models have been used for face recognition and among them probabilistic methods However, up to now probabilistic methods rely heavily on the number of training data and do not fully exploit the 2-dimensional information of the images, both the training and the testing sets. In this paper’s method a new 2-D robust probabilistic method of transforming the principal components of the initial image data, allowing support vector machines to efficiently capture the inference between images. This new algorithm encodes every image with the help of Robust Kernel non Parametric Estimation and in the second stage uses Support Vector Machines to classify this encoded information. Results exhibit that Non Parametric Estimation of the Probability Function of the image highlights the unique characteristics of each person making it easier for classifiers to group those instances and efficiently perform the classification of the images and thus leading to better results compared to up to date methods for face recognition.

Paper Details

Date Published: 16 April 2014
PDF: 6 pages
Proc. SPIE 9159, Sixth International Conference on Digital Image Processing (ICDIP 2014), 915926 (16 April 2014); doi: 10.1117/12.2064730
Show Author Affiliations
Meropi Pavlidou, Aristotle Univ. of Thessaloniki (Greece)
George Zioutas, Aristotle Univ. of Thessaloniki (Greece)

Published in SPIE Proceedings Vol. 9159:
Sixth International Conference on Digital Image Processing (ICDIP 2014)
Charles M. Falco; Chin-Chen Chang; Xudong Jiang, Editor(s)

© SPIE. Terms of Use
Back to Top