Face Recognition using Tridiagonal Matrix Enhanced Multivariance Products Representation
Küçük Resim Yok
Tarih
2017
Yazarlar
Dergi Başlığı
Dergi ISSN
Cilt Başlığı
Yayıncı
Amer Inst Physics
Erişim Hakkı
info:eu-repo/semantics/closedAccess
Özet
This study aims to retrieve face images from a database according to a target face image. For this purpose, Tridiagonal Matrix Enhanced Multivariance Products Representation (TMEMPR) is taken into consideration. TMEMPR is a recursive algorithm based on Enhanced Multivariance Products Representation (EMPR). TMEMPR decomposes a matrix into three components which are a matrix of left support terms, a tridiagonal matrix of weight parameters for each recursion, and a matrix of right support terms, respectively. In this sense, there is an analogy between Singular Value Decomposition (SVD) and TMEMPR. However TMEMPR is a more flexible algorithm since its initial support terms (or vectors) can be chosen as desired. Low computational complexity is another advantage of TMEMPR because the algorithm has been constructed with recursions of certain arithmetic operations without requiring any iteration. The algorithm has been trained and tested with ORL face image database with 400 different grayscale images of 40 different people. TMEMPR's performance has been compared with SVD's performance as a result.
Açıklama
11th International Conference on Mathematical Problems in Engineering, Aerospace and Sciences (ICNPAA) -- JUL 04-08, 2016 -- Univ La Rochelle, La Rochelle, FRANCE
Anahtar Kelimeler
Kaynak
Icnpaa 2016 World Congress: 11th International Conference On Mathematical Problems In Engineering, Aerospace And Sciences
WoS Q Değeri
N/A
Scopus Q Değeri
N/A
Cilt
1798
