R. Koch. Dynamic 3D scene analysis through synthesis feedback control. IEEE Transactions on Pattern Analysis and Machine Intelligence, 15(6):346-- 351, July 1993.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....faces can conversational interfaces [5] interactive kiosks [24] and robots [11] start to understand more features of natural face to face dialogue. And only by gathering accurate motion estimates of faces can facial animation systems be automated [27] or videos of faces be compressed effectively [15, 16, 27]. Even though some of these systems do not need to know about the user s appearance, having an accurate estimate of the face shape is still important, # D. DeCarlo is with the Department of Computer Science and Center for Cognitive Science, Rutgers University, New Brunswick, NJ. E mail: ....

....and motion parameters as the adjustment is computed. For example, it is insufficient to simply stage the computation, and use the leftovers from the motion estimate to feed a computation which determines how the shape parameters could have changed over time to explain the remaining observed motion [15]. This method simply treats shape parameters as motion parameters. In this paper, we propose computing an adjustment to the shape parameters that minimizes the error in the motion estimate. In other words, we determine a new configuration for which the motion parameters would have produced less ....

[Article contains additional citation context not shown here]

R. Koch. Dynamic 3-D scene analysis through synthesis feedback control. IEEE Pattern Analysis and Machine Intelligence, 15(6):556--568, June 1993.

....of the resulting 3D motion compensation method. Alternatively, 3D models of objects may be derived from stereo images. This usually requires estimation of dense disparity elds, postprocessing to remove erroneous estimates and tting of a parametrised surface model to the calculated depth map [14]. In [17] an algorithm was presented which optimally models the scenes using a hierarchically structured wire frame model derived directly from intensity images. The wire frame model consists of adjacent triangles that may be split into smaller ones over areas that need to be represented in higher ....

R. Koch, Dynamic 3D scene analysis through synthesis feedback control, IEEE Trans. Pattern Anal. and Mach. Intell. 15 (June 1993) 556}568.

....the model does not have an extremely large set of motion parameters. This would preclude the use of model based techniques over image based techniques. 19 Aside from tracking, it is also possible to use a model based optical flow formulation to estimate the model structure. In particular, Koch [Koc93] describes a model based framework which uses optical flow information to estimate the rigid translation and rotation of a moving face, and adapts the shape of the face to account for the motion discrepancy. Chapter 6 presents an alternative method of structure estimation from optical flow ....

....Edge information is not always adequate due to poor illumination and self occlusion. This may result in inaccurate estimation of the basic shape, which can in turn cause error in the motion estimation. This approach also differs from other model based shape and motion estimation methods [Koc93] where optical flow information was used to directly improve the shape, leading to potentially large shape estimation errors. Our method does not require the extraction of tracked features, but instead uses motion information in this case, optical flow information. Shape and motion are improved ....

[Article contains additional citation context not shown here]

R. Koch. Dynamic 3-D scene analysis through synthesis feedback control. IEEE Pattern Analysis and Machine Intelligence, 15(6):556--568, June 1993.

....elastic properties of real materials for the recovery of structure and nonrigid motion. Huang and Goldgof [8] proposed the adaptive size physically based models for nonrigid shape and motion analysis. Many applications of model based structure and nonrigid motion recovery were also presented [9, 11, 17, 18, 7, 3, 12]. For example, Kakadiaris and Metaxas [9] discussed the human body tracking, DeCarlo and Metaxas [7] estimated the shape and motion of human faces with a deformable model. However, the major limitation of such methods is that only overconstrained global shape descriptions of nonrigid motion are ....

R. Koch. Dynamic 3-d scene analysis through synthesis feedback control. PAMI, 15(6):556--568, June 1993.

....synthesized facial expressions. 3.3 Facial Parameter Estimation Our FAP estimation algorithm estimates the facial animation parameters from two successive frames of a natural or rendered video sequence. To avoid error accumulation in the long term parameter estimation, a feedback loop is used [20,21] as depicted in Fig. 2. The model of the head is moved according to the Model Analysis Synthesis Shape Texture Expressions model update facial parameters input I(t) t 1) Fig. 2. Feedback structure of the coder. parameters estimated from video frames I(t) and I(t 1) and a synthetic image is ....

R. Koch. Dynamic 3-D scene analysis through synthesis feedback control. IEEE Transactions on Pattern Analysis and Machine Intelligence, 15(6):556-- 568, Jun. 1993.

.... mesh that represents the shape of the objects [24] The texture of the objects is initialised by mapping parts of the image triple onto each triangle [18] The update phase starts by analysing the motion of the objects, assuming that most of the changes result from motion of rigid objects [12]. In a next step, the shape of the 3 D models is updated. This update considers flexible deformations of the object which cannot be described by motion and improves the shape over time. Finally the texture of the 3 D models is updated, compensating the changes of the objects surface colour. The ....

R. Koch, Dynamic 3-D Scene Analysis through Synthesis Feedback Control, IEEE PAMI, Vol. 15, Nr. 6, S556-568, Juni 1993.

....This implies the need to estimate the camera positions for all view points. 4. 1 Estimation of 3 D Camera View Point The camera position can be derived directly from the spatial and temporal image gradients as long as the relative camera motion is small between consecutive image frames [6] [7]. It is computed by tracking the relative motion of the 3 D objects visible to the camera and then adjusting the camera position accordingly. An object is defined as a rigid 3 D surface in space that is spanned by a set of N control points. Six motion parameters are associated with the object. ....

Koch, R. Dynamic 3D Scene Analysis through Synthesis Feedback Control. IEEE Trans. Patt. Anal. Mach. Intell., Special issue on analysis and synthesis 1993. Vol. 15(6):556--568.

No context found.

R. Koch. Dynamic 3D scene analysis through synthesis feedback control. IEEE Transactions on Pattern Analysis and Machine Intelligence, 15(6):346-- 351, July 1993.

No context found.

R. Koch. Dynamic 3D scene analysis through synthesis feedback control. IEEE Trans. Pattern Analysis and Machine Intelligence, 15(6):556--568, 1993.

No context found.

R. Koch. Dynamic 3-D scene analysis through synthesis feedback control. IEEE Transactions on Pattern Analysis and Machine Intelligence, 15(6):556-- 568, Jun. 1993.

No context found.

R. Koch. Dynamic 3D scene analysis through synthesis feedback control. IEEE Transactions on Pattern Analysis and Machine Intelligence, 15(6):556--568, June 1993.

No context found.

R. M. Koch. Dynamic 3D scene analysis through synthesis feedback control. IEEE Transactions on Pattern Analysis and Machine Intelligence, 15(6):556-- 568, June 1993.

No context found.

R. M. Koch. Dynamic 3D scene analysis through synthesis feedback control. IEEE Transactions on Pattern Analysis and Machine Intelligence, 15(6):556-- 568, June 1993.

No context found.

Koch R. Dynamic 3-D scene analysis through synthesis feedback control IEEE T-PAMI 15, 1993, p.556-568.

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