| K. Sobottka and I. Pitas, "Localization of Facial Regions and Features in Color Images", 4th Open Russian-German Workshop: Pattern Recognition and Image Analysis, Valday, The Russian Federation, pp. 134-138, 3-9 March 1996 |
....searching time and prevents skin color in background or clothing or hands, to be added to face color region. Then we apply a connected component algorithm to the segmented area and compute position and orientation of face by the computation of moments of face area. Our approach is similar to [14]. The face size is used to control the zoom of camera. 6. Tracking In general, there are two approaches for tracking that are fundamentally different [3] These are recognition based tracking and motion based tracking. In recognition based tracking, the object is recognized in successive images ....
K. Sobottka and I. Pitas, "Localization of Facial Regions and Features in Color Images", 4th Open Russian-German Workshop: Pattern Recognition and Image Analysis, Valday, The Russian Federation, pp. 134-138, 3-9 March 1996
....the x and y directions and I k;l is the intensity value of the pixel (k; l) The l direction corresponds to that of the main axis of an ellipse fitting the spatial distribution of the pixels belonging to the face skin mask. The best fit ellipse is estimated by the method of moments (see e.g. [20]) Then the values of the magnitude of the computed directional derivative are projected onto the ellipse major axis: P (x; y) X k D[ x; y) kn] 2) where n is the normal to the main ellipse axis, i.e. the minor axis orientation. 2.3 Selecting the best candidate The lips will lie around a ....
K. Sobottka and I. Pitas. Localization of facial regions and features in color images. Journal of Pattern Recognition and Image Analysis, 1996.
....the 5 Theta 3 rectangle. Results of this preprocessing step are illustrated in Figure 3b. Eyes and mouth and parts of the hair and beard are emphasized. The positions of eyes and mouth are determined by searching for minima in the topographic greylevel relief. This can be done by using watersheds [9], or as we will show here, by directly evaluating the x and y projections of the greylevel relief (Figure 4) Since eyes and mouth are horizontally orientated, a normalization of the orientation of the face candidate is necessary. The advantage of this normalization step is that, afterwards, we ....
K. Sobottka and I. Pitas. Localization of facial regions and features in color images. In 4th Open RussianGerman Workshop: Pattern Recognition and Image Analysis, pages 134--138, Valday, The Russian Federation, March 3-9 1996.
No context found.
K. Sobottka and I. Pitas. Localization of facial regions and features in color images. In 4th Open Russian-German Workshop: Pattern Recognition and Image Analysis, pages 134--138, Valday, The Russian Federation, March 3-9 1996.
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