Part-Based And Multispace Random Mapping For Face Recognition
In this thesis, we studied several part-based approaches for face recognition. Our objective is to reduce the dimensionality of the raw image while retaining as many salient features as possible. Non-Negative Matrix Factorization (NMF) is recently a proposed method to obtain a part-based linear repr...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
2005
|
| Subjects: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | In this thesis, we studied several part-based approaches for face recognition. Our objective is to reduce the dimensionality of the raw image while retaining as many salient features as possible. Non-Negative Matrix Factorization (NMF) is recently a proposed method to obtain a part-based linear representation of facial image. However, the bases learned by NMF do not display perfectly the local characteristics as there are still some non-zero weight values in the features. These values appear as noise and contribute to the degradation of the recognition performance. In this thesis, we have proposed a novel part-based feature extractor based on NMF. |
|---|