D. Reisfeld, Generalized Symmetry Transforms: Attentional Mechanisms and Face Recognition, Ph.D. dissertation, Tel Aviv Univ., Jan. 1994.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....by the value s f . This leads to a set of invariant features. In addition, we note that there is also a particular spatial relationship between the eyes, mouth, and nose which is governed by the scale s f . This spatial relationship for example the triangle formed by the two eyes and the mouth [13] can be also be used to normalize facial images and results in another set of invariant features. 36 4. Results We present a set of results that have been obtained by employing the methods discussed in Sections 2 and 3 to locate the most perceptually significant parts and to construct ....

D. Reisfeld. Generalized Symmetry Transforms: Attentional Mechanisms and Face Recognition. PhD thesis, Tel-Aviv University, Tel-Aviv, Israel, 1994.

....To thisissue, we utilize a class of Figure 3. Initialization (first column) and results obtained with 1 (second column) and then 3 (third column) harmonics for the upper boundary of mouth and eyes and the lower boundary of nose. deformations similar to the one used for normalization by Reisfeld [10]. While his aim was to normalize faces in order to apply a classification on the image data, we propose to achieve the identification using the deformation measure value as the discriminating information. 4.1. Face characterization and database The landmark points used to represent the face are ....

....T . In particular the use of U(x) x 2 log(x) with U(0) 0) leads to an unique solution verifying the property of the thin plate spline. The coefficients a 1 ; a 2 ; a 3 ; b 1 ; b 2 ; b 3 ; w 1 ; wN and z 1 ; z N can be calculated by solving a linear system as described in [2] [10]. The quantity J(T ) may be used to measure the likeness between two faces. The assumed identity will then be confirmed if this value is smaller than a given threshold. The threshold value is linked to the accuracy of the system that is wished for. If it is small, it can hapen that the identity of ....

D. Reisfeld. Generalized symmetry transforms: attentional mechanisms and face recognition. PhD thesis, Tel-Aviv University, January 1994.

....[19] Let us consider rst the position of a skewed symmetry edge. Since a midpoint of a segment is an aOEne invariant, the position is the same as in the previous case. The de nition of a skewed symmetry edge orientation is, however, dioeerent. It follows from basic trigonometric considerations [28] that the natural skewed symmetry edge orientation is (i; j) arctan 2 cos( i Gamma ff i;j ) cos( j Gamma ff i;j ) sin( i j Gamma 2ff ij ) where ff ij is de ned as in section 3. One may need to add to be consistent with the ideal case. 7 Acknowledgements We thank M. D. Levine and ....

D. Reisfeld. Generalized symmetry transforms: attentional mechanisms and face recognition. PhD thesis, Computer Science Department, Tel-Aviv University, January 1994.

....Let us consider first the position of a skewed symmetry edge. Since a midpoint of a segment is an affine invariant, the position is the same as in the previous case. The definition of a skewed symmetry edge orientation is, however, different. It follows from basic trigonometric considerations [28] that the natural skewed symmetry edge orientation is (i; j) arctan 2 cos( i Gamma ff i;j ) cos( j Gamma ff i;j ) sin( i j Gamma 2ff ij ) where ff ij is defined as in section 3. One may need to add to be consistent with the ideal case. 7 Acknowledgements We thank M. D. Levine and ....

D. Reisfeld. Generalized symmetry transforms: attentional mechanisms and face recognition. PhD thesis, Computer Science Department, Tel-Aviv University, January 1994.

No context found.

D. Reisfeld, Generalized Symmetry Transforms: Attentional Mechanisms and Face Recognition, Ph.D. dissertation, Tel Aviv Univ., Jan. 1994.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC