| M. Leonidou, N. Tsapatsoulis and S. Kollias |
| Face Recognition based on Multiple Representations: Splitting the Error Space |
| Advances in Intelligent Systems, Concepts Tools and Applications, PART III: Image Processing and Video Based Systems, Kluwer Academic Publishers, 1999 |
|
ABSTRACT
|
| The proposed, in this paper, scheme aims at recognizing faces irrespectively of position, orientation, scale, luminance, and texture variances in a different approach. We claim that the selection of multiple and mutually exclusive representations can overcome the difficulty of finding a global set of features. Multiple and complementary representations are adopted so as to split the error domain (the union of orientation, scale, luminance and texture variances) leading to successful parallel partial solutions for the different disciplines of face recognition problems. This is accomplished by embedding each representation with a suitable identification algorithm. In a subsequent stage a rule-based decision mechanism selects the most reliable solution, taking into account the partial identification results and an estimation of the variance that caused the difference between the test face and its stored version, if any.
|
| 15 May , 1999 |
| M. Leonidou, N. Tsapatsoulis and S. Kollias, "Face Recognition based on Multiple Representations: Splitting the Error Space", Advances in Intelligent Systems, Concepts Tools and Applications, PART III: Image Processing and Video Based Systems, Kluwer Academic Publishers, 1999 |
[
BibTex] [
Print] [
 Back] |
