Y. Li, S. Gong, and H. Liddell. Modelling faces dynamically across views and over time. In Proc. IEEE International Conference on Computer Vision, 1:554--559, 2001.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....module. The most reliable video frames and audio clips are selected for recognition. 3D information about the head is used to detect the presence of an actual person as opposed to an image of that person. Recognition and veri cation rates of 100 were achieved for 26 registered clients. In [24], recognition of face over time is implemented by constructing a face identity surface. The face is rst warped to a frontal view, and its Kernel Discriminant Analysis (KDA) features over time form a trajectory. It is shown that the trajectory distances accumulate recognition evidence over time. ....

Y. Li, S. Gong, H. Liddell, Modelling faces dynamically across views and over time, Proc. of ICCV (2001) 554-559.

....in many computer vision and pattern recognition applications. Active Shape Models (ASM) 6, 3] and Active Appearance Models (AAM) 4] proposed by Cootes et al., are two popular models for the purpose of shape and appearance modeling and extraction. They have received much attention in recent years [10, 1, 15, 16, 2, 12, 11, 9, 18, 14, 7, 8]. In ASM, the local appearance model, which represents the local statistics around each landmark, e#ciently finds the best candidate point for each landmark in searching the image. The solution space is constrained by a properly trained global shape model. Based on the accurate modeling of the ....

Y. Li, S. Gong, and H. Liddell. "Modelling faces dynamically across views and over time". In Proceedings of IEEE International Conference on Computer Vision, pages 554--559, Vancouver, Canada, July 2001.

No context found.

Y. Li, S. Gong, H. Liddell, Modelling faces dynamically across views and over time, in: IEEE International Conference on Computer Vision, Vancouver, Canada, 2001, pp. 554 -- 559.

No context found.

Y. Li, S. Gong, and H. Liddell. Modelling faces dynamically across views and over time. In IEEE International Conference on Computer Vision, pages 554--559, Vancouver, Canada, July 2001.

....[ 40, 40] in yaw with an interval of 10 ) 13] Each image in the training set is associated with known pose in tilt and yaw and labelled semi automatically with a set of salient landmarks including the positions of eyes, nose, mouth and face contour. More details of the model are discussed in [18]. Figure 2 shows the original face images, fitted multi view face model overlaid on the face images, and the warped shape and pose free texture patterns. Figure 2: Extract the shape and pose free texture patterns of multi view face images using a multi view dynamic face model. When one side of ....

....tilt and 40 40 in yaw with an interval of 10 ) and the synthesised identity surface using only 15 example views (same ranges but with an interval of 20 ) 4. 3 Recognition by Pattern Distances to the Identity Surfaces For an unknown face image, one first fits the multi view dynamical face model [18] onto the image and projects onto the KDA feature space to yield a face pattern (x, y, z0) where z0 is the KDA vector and x, y are the pose in tilt and yaw, then the pattern distance to one of the identity surfaces can be computed as the Euclidean distance between z0 and the corresponding point z ....

[Article contains additional citation context not shown here]

Y. Li, S. Gong, and H. Liddell. Modelling faces dynamically across views and over time. Technical report, Queen Mary, University of London, 2001. www.dcs.qmw.ac.uk/oyongmin/papers/model.ps.

....from images are required in many computer vision and pattern recognition applications. Active Shape Models (ASM) and Active Appearance Models (AAM) proposed by Cootes et al. 4] are two popular shape and appearance models for object localization. They have been developed and improved for years [5 7,9]. In ASM, the local appearance model, which represents the local statistics around each landmark, efficiently finds the best candidate point for each landmark in searching the image. The solution space is constrained by the properly trained global shape model. By means of modeling of the local ....

Y. Li, S. Gong, H. Liddell, Modelling faces dynamically across views and over time, Proceedings of the IEEE International Conference on Computer Vision, Canada July (2001) 554 -- 559.

....The multi view dynamic face model is described in Section 2. Identity surface synthesis, object and model trajectory construction, and dynamic face recognition are presented in Section 3. Conclusions are drawn in Section 4. 2 Multi View Dynamic Face Model Our multi view dynamic face model [12] consists of a sparse 3D Point Distribution Model (PDM) 4] learnt from 2D images in different views, a shape and pose free texture model, and an affine geometrical model which controls the rotation, scale and translation of faces. The 3D shape vector of a face is estimated from a set of 2D face ....

....face pattern is # shape parameter, is the texture parameter, is pose in tilt and yaw, is the translation of the centroid of the face, and is its scale. More details of model construction and fitting are described in [12]. The shape and pose free texture patterns obtained from model fitting are adopted for face recognition. In our experiments, we also tried to use the shape patterns for recognition, however, the performance was not as good as that of using textures. 3 Recognising Faces Using Identity Surfaces ....

Y. Li, S. Gong, and H. Liddell. Modelling faces dynamically across views and over time. In IEEE International Conference on Computer Vision, Vancouver, Canada, July 2001.

....is f 3 P haPQi P j P3k K PQkalmPQno 1 where g is the shape parameter, h is the texture parameter, ipPQj, is pose in tilt and yaw, K P3kqlr is the translation of the centroid of the face, and n is its scale. More details of model construction and fitting are described in [11]. The shape and pose free texture patterns obtained from model fitting are adopted for face recognition. In our experiments, we also tried to use the shape patterns for recognition, however, the performance was not as good as that of using textures. 4 Extracting the Non linear Discriminating ....

Y. Li, S. Gong, and H. Liddell. Modelling faces dynamically across views and over time. In IEEE International Conference on Computer Vision, Vancouver, Canada, July 2001.

No context found.

Y. Li, S. Gong, and H. Liddell. Modelling faces dynamically across views and over time. In Proc. IEEE International Conference on Computer Vision, 1:554--559, 2001.

No context found.

Y. Li, S. Gong, and H. Liddell. Modelling faces dynamically across views and over time. ICCV, 1:554--559, 2001.

No context found.

Y.M. Li, S.G. Gong, and H. Liddell. Modelling faces dynamically across views and over time. In Proc. of International Conf. on Computer Vision, pages I: 554--559, 2001.

No context found.

Y. Li, S. Gong, H. Liddell, Modelling faces dynamically across views and over time, Proc. ICCV (2001) 554--559.

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