K. Sobottka and I. Pitas, "Extraction of facial regions and features using color and shape information," in Proceedings of the International Conference Of Pattern Recognition (ICPR '96), pp. 421--425, August 1996.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....measure, denoted as ) b a L v , accounts for the textural properties of skin regions, and therefore significantly reduces the false acceptance in detecting the skin candidates. Fig. 4: Examples of skin color detection using (from the left to the right) the methods proposed in [18] in [19] and in this section with the variables (L, a, b) and, in last column, with the variables (L, a, b, v(L, a, b) Notice how the additional variable, related to texture properties, enhances the classification in images that do not contain skin. 8. NATURAL QUERY LANGUAGE This section presents a ....

K. Sobottka and I. Pitas, "Extraction of facial regions and features using color and shape information", International Conference on Pattern Recognition (ICPR'96), Vienna, Austria, vol. III, pp. C421-C425, 25-29 August 1996.

....of white and black, and uses these values to compute a modified Von Kries adaptation. Although the spectrum of the light source cannot be completely recovered from the image, this model provides good results, as long as the spectrum of the light source is not too wildly skewed or irregular [28]. Fig. 5 shows a comparison of our detection method with two others on both images containing skin and images Fig. 2. The origin of this real data set is not mentioned for now otherwise it would clearly identify the authors. On the left, a cloud of points in R3 distributed along a one dimensional ....

K. Sobottka and I. Pitas. Extraction of facial regions and features using color and shape information. In Proceedings of the International Conference on Pattern Recognition, volume 3, pages C421 C425, Vienna, Austria, 1996.

....a set of local feature detectors via a statistical model to find the facial components for face locating. Their apprach was invariant with respect to translation, rotation, and scale. Besides, they could also handle partial occlusion of faces. Instead of the gray scale images, Sobottka and Pitas[23], and Chen et al. 24, 25, 26, 27, 28] located the poses of human faces and facial 3 features from color images. In [23] the oval shape of a face could be approximated by an ellipse in Hue Saturation Value(HSV) color space. Chen et al. 24, 25, 26, 27, 28] proposed a skin color distribution ....

....was invariant with respect to translation, rotation, and scale. Besides, they could also handle partial occlusion of faces. Instead of the gray scale images, Sobottka and Pitas[23] and Chen et al. 24, 25, 26, 27, 28] located the poses of human faces and facial 3 features from color images. In [23], the oval shape of a face could be approximated by an ellipse in Hue Saturation Value(HSV) color space. Chen et al. 24, 25, 26, 27, 28] proposed a skin color distribution function on perceptually uniform color space to detect the face like region. The skin color regions in color images were ....

K. Sobottka and I. Pitas, "Extraction of facial regions and features using color and shape information", in Proc. 13th International Conference on Pattern Recognition, Vienna, Austria, Aug. 1996, pp. 421--425.

....face identi cation, model based video coding and facial expression recognition, as most algorithms require that the faces be rst detected and localized. Quite a few face detection studies have been reported in the literature. Some use monochromatic images [8, 4, 6] and some use color images [1, 3, 7]. Many assume that the faces to be detected are in the frontal poses [8, 4, 6] and that the sizes of faces to be detected are within a certain range [4, 6, 1, 7] Some use pattern matching, with the patterns being prepared manually or by learning[6, 2, 1, 7, 5] Since the face size and position ....

....use a 2D Gaussian for skin color modelling and achieve comparable speed for tracking. Though we have not done any experiments on faces of other races than East Asians, the work reported by Yang et al. and Sobottka et al. shows that similar approaches should work for people with other skin colors [9, 3]. 2. Skin Color Modeling and Detection In our approach, we use RGB color space and model the distribution of skin color within a particular race as a 3D Gaussian distribution, enabling segmentation of skin color from non skin color by the Mahalanobis distance from the center of the Gaussian ....

K. Sobottka and I. Pitas. Extraction of facial region and features using color and shape information. In Proc. 10th Int'l Conf. Pattern Recog., pages C421-425, Vienna, 1996.

....variations. One solution, adopted in [32] is that each person should use his her own colour information, i e the colour information of the specific person should be known in advance. In addition, the classification should be calibrated for the specific camera used. One approach is presented in [33], where shape as well as colour information is used. First, skin like regions are located by performing colour segmentation. Then, regions of nearly elliptical shape are searched for. A more advanced method can be found in [34] using a likelihood ratio of skin colour (instead of a binary ....

K. Sobottka and I. Pitas, "Extraction of facial regions and features using color and shape information," Proc. 13th IAPR, pp. 421--425, 1996.

.... is difficult to automatically detect human heads or faces in images having complex backgrounds, much previous research either bypassed the problem of human head detection by manually locating the head in the image before tracking [1] 6] or dealt only with images having simple backgrounds [3] 9] 15][17]. However, for many practical applications, detection and tracking of human heads have to be automatic (i.e. no manual initialization is allowed) and should not be limited to simple backgrounds. The approaches to head (or face) detection can be model based [9] 15] 17] 18] feature based [2] 19] ....

....simple backgrounds [3] 9] 15] 17] However, for many practical applications, detection and tracking of human heads have to be automatic (i.e. no manual initialization is allowed) and should not be limited to simple backgrounds. The approaches to head (or face) detection can be model based [9] 15][17][18] feature based [2] 19] neural network based [13] 14] or color based [15] 17] In the model based approach, several researchers modelled the shape of a human head with an ellipse, and then used ellipse fitting methods to locate the human head [9] 15] 17] This work was partially ....

[Article contains additional citation context not shown here]

K. Sobottka and I. Pitas, "Extraction of Facial Regions and Features Using Color and Shape Information", Proc. of ICPR'96, Vol. 3, pp. 421-425.

....of faces is given in [3] Also [1] gives a very good overview about existing approaches and techniques in this research field. The present paper proposes a novel approach for facial feature extraction. The segmentation of facial regions is done in advance by using color and shape information ([8], 9] We employ eyebrows, eyes, nostrils, mouth and chin as interesting facial features. Facial features are localized by evaluating the topographic greylevel relief of facial regions. Once facial features are reliably detected, we track them over time. Tracking is performed by searching for ....

K. Sobottka and I. Pitas. Extraction of facial regions and features using color and shape information. In Int. Conf. on Pattern Recognition (ICIP), Vienna, Austria, August 1996.

No context found.

K. Sobottka and I. Pitas, "Extraction of facial regions and features using color and shape information," in Proceedings of the International Conference Of Pattern Recognition (ICPR '96), pp. 421--425, August 1996.

No context found.

K. Sobottka and I. Pitas, "Extraction of facial regions and features using color and shape information," in Proc. ICIP 96, vol. 3, pp. 483--486, 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