A Feature Based Algorithm for Face Image Description

• Published in: 2011 International Conference on Biometrics and Kansei Engineering pp. 175 - 178
• Main Authors: Kocjan, Przemyslaw, Saeed, Khalid
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.09.2011
• Subjects: biometric authentication; Biometrics; Eigenvalues and eigenfunctions; Face; Face recognition; Humans; Mathematical model; Toeplitz matrices
• ISBN: 145771356X; 9781457713569
• DOI: 10.1109/ICBAKE.2011.17

The process of automatic identification or verification of the human face is a very interesting problem in both sociology and industry. The way of gathering biometric information is non intrusive and easy to perform. Systems that can scan faces in big public places like the airport in search of the offender's face are becoming more and more popular. This paper studies the use of the modified Toeplitz matrices as a face descriptor and shows results of classification using classifiers like kNN, Naive Bayes and Discriminant Analysis. Many face recognition algorithms are based on statistical approach like PCA, LDA, ICA and their modifications, other statistical approaches include ASM, AAM and others. Points obtained form i.e. ASM may be used as an input to create the feature vectors of the face image. The proposed way of the face description was successfully used in written text recognition but it was never shown as a fully successful descriptor in the face identification purposes.