Face Recognition using Local Features based on Two-layer Block Model.
Wonjun Hwang, Jiyeon Kim, Seok-Cheol Kee · 2005
This paper proposes the novel face recognition algorithm, local features basedon two-layer blockmodel, in order to improve the generalization abilityof the face recognition. The traditional LDA scheme is often unstable even though it is the popular extraction technique for face recognition. In this paper, we focus on the performance stabilitywhen we have tests whose property is different from trained variations.Local Feature Analysis is adoptedto transfer a face image into several local block representations by different blockmodels, andLinear Discriminant Analysis is usedto increase the discriminant power of separatedblock representations.The methodwas testedon three different face database andthe system was shown to perform very well when comparedto traditional approach.