Browsing by Author "Tesfay, Tamirat"

Now showing 1 - 2 of 2
  • No Thumbnail Available
    Item
    Impact of Face Image Currency on Recongition Rates of the Eigenface Algorithm
    (Addis Ababa University, 2005-06) Tesfay, Tamirat; Ricanek, Karl(PhD)
    A technology evaluation is administrated to assess the impact of face image currency on the recognition rates of the Eigenface algorithm. Several experiments have been carried out to highlight the challenging problem in Face Recognition, -- ageing or aging. The MORPH database is used as a data source since it contains long-term data that spans from few days up to 29 years, between the acquisition of the first (gallery) and the subsequent (probe) image sets. The Identification test indicates that 1) the performance of the Eigenface algorithm decreased linearly (in general) with age-progression. 2) The Eigenface algorithm performed better in identifying older people than younger people. 3) The algorithm performed better in identifying males at younger age than females. 4) The algorithm achieved more or less similar results for the African-American and the Caucasian ethnic origin. In general, the overall performance evaluation test on the Eigenface algorithm indicates that the performance of the Eigenface approach is very sensitive to age-progression. Keywords: Eigenface, age-progression, MORPH, FERET, FRVT, PCA
  • No Thumbnail Available
    Item
    Impact of Face Image Currency on Recongition Rates of the Eigenface Algorithm
    (Addis Ababa University, 2005-06) Tesfay, Tamirat; Ricanek, Karl(PhD)
    A technology evaluation is administrated to assess the impact of face image currency on the recognition rates of the Eigenface algorithm. Several experiments have been carried out to highlight the challenging problem in Face Recognition, -- ageing or aging. The MORPH database is used as a data source since it contains long-term data that spans from few days up to 29 years, between the acquisition of the first (gallery) and the subsequent (probe) image sets. The Identification test indicates that 1) the performance of the Eigenface algorithm decreased linearly (in general) with age-progression. 2) The Eigenface algorithm performed better in identifying older people than younger people. 3) The algorithm performed better in identifying males at younger age than females. 4) The algorithm achieved more or less similar results for the African-American and the Caucasian ethnic origin. In general, the overall performance evaluation test on the Eigenface algorithm indicates that the performance of the Eigenface approach is very sensitive to age-progression. Keywords: Eigenface, age-progression, MORPH, FERET, FRVT, PCA