E. Dubois, "Motion-compensated filtering of time-varying images," Multidimensional Syst. Signal Process., vol. 3, pp. 211--239, 1992.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....Motion within a sequence of images and the displacement vector are defined. Concepts concerning partitioning the image into blocks of variable sizes are introduced. Finally, the formulation of the objective function is discussed. The theoretical formulation extends the one given by Dubois[2], Konrad[8] Bergeron[25] Juillet[11] and Depommier[6] hence, the terminology used is the same. 2.1 Motion in a Image Sequence Motion in a scene is captured on a sequence of images. At any instant in time, the threedimensional scene is projected on a two dimensional image surface. With a ....

....methods for motion estimation involve minimization of an objective function consisting of two terms: structural and smoothness. This is due to the random nature of the observed images and the difficulties associated with computing motion. The formulation of the objective function as given by Dubois[2] is U(d) S(d) P (d) 2.7) where S(d) corresponds to the structural term and P (d) corresponds to the smoothness term. 2.5.1 Structural term The structural term refers to a local feature of the motion field. We consider a part of the image and compute the motion vector that maximizes the ....

[Article contains additional citation context not shown here]

Dubois E., Motion Compensated Filtering of Time-Varying Images, Multidimensional Systems and Signal Processing, vol. 3, 1992, pp. 103-131.

....of uncovered covered backgrounds, and the presence of sensor and coding noise further complicate the problem. A discussion on spatio temporal sampling, relationships between the spatial resolution and frame rate, as well a discussion of motion compensated filtering is presented by Dubois in [64]. He also describes interpolative coding methods that are based on dropping fields at the encoder and reconstructing the missing fields at the decoder using interpolation. Bierling and Thoma [25] describe a field interpolation algorithm. They used a hierarchically structured motion estimator, ....

E. Dubois, "Motion-compensated filtering of time-varying images," Optical Engineering, vol. 3, pp. 211--239, 1992.

....comply with real time requirements and bandwidth constraints. In these situations, it would be advantageous to provide the decoder with the option of increasing the frame rate. Such applications involve a resampling of the spatio temporal sampling grid, and processing along motion trajectories [5]. Thus, when motion estimates are accurate and close to the true motion, we can expect the video representation (motion plus residual) to scale better with temporal or spatial resolution. Furthermore, accurate motion estimates may be expected to improve compression performance. However, most ....

....presence of uncovered covered backgrounds, and the presence of sensor and coding noise further complicates the problem. A discussion on spatio temporal sampling, relationships between spatial resolution and frame rate, as well a discussion of motion compensated filtering is presented by Dubois in [5]. He also describes interpolative coding methods that are based on dropping fields at the encoder and reconstructing the missing frames at the decoder using interpolation. Other references that describe coders with field skipping or frame skipping are presented by Bierling and Thoma [11, 12] ....

E. Dubois, "Motion-compensated filtering of time-varying images," Optical Engineering, vol. 3, pp. 211--239, 1992.

.... preferred texture orientation in the image, proved advantageous in terms of visual rendition [1, 2, 3, 4, 5, 6] Velocity may be interpreted as orientation in spatio temporal domain, and motion compensated spatio temporal filters may be used succesfully for prediction, interpolation and smoothing [7] as well as for coding [8] Regardless of the specific descriptor of interest, most techniques start processing the image (or image sequence) with a family of linear filters tuned at a wide range of orientations and scales of resolution [9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21] The ....

E. Dubois. Motion--compensated filtering of time--varying images. Multidimensional Systems and Signal Processing, 3:211--239, 1992.

....Science Foundation Engineering Research Center Program; and by the California Trade and Commerce Agency, Office of Strategic Technology. The author R. Manduchi is currently with Apple Computer, Inc. Cupertino, CA. filters may be used succesfully for prediction, interpolation and smoothing [2] as well as for coding [3] Regardless of the specific descriptor of interest, most techniques start processing the image (or image sequence) with a family of linear filters tuned at a wide range of orientations and scales of resolution (see [4] for an extensive bibliography on the subject) The ....

E. Dubois. Motion--compensated filtering of time--varying images. Multidimensional Systems and Signal Processing, 3:211--239, 1992.

....Section 3.5. Finally, experimental results of the proposed algorithm are presented in Section 3.6 along with a comparison between the linear and quadratic motion trajectory models. 3. 1 Definition of motion trajectory To describe motion with acceleration, the concept of motion trajectory is used [11]. The projection of each scene point traces out a trajectory in the image plane W during the time it is visible in the image. Hence, the motion in the image sequence is characterized by the collection of all such trajectories. An illustration of a typical trajectory of the center of a circle ....

E. Dubois, "Motion-Compensated Filtering of Time-Varying Images," Multidimensional Systems and Signal Processing, vol. 3, pp. 211--239, 1992.

....vary in time t, it is useful to consider the trajectory of an image point in a conceptual 3 D xyt space. Let the function c( x; t) mathematically describe a trajectory in the image plane, i.e. let c( x; t) be the spatial position at time of an image point which at time t was located at x [14]. c( x; t) describes a 2 D trajectory in the image plane, while (c( x; t) describes a 3 D trajectory in the xyt space. Clearly, there is a unique mapping between the two trajectories. The shape of the trajectory c( x; t) depends on the nature of object motion. We define the ....

E. Dubois, "Motion-compensated filtering of time-varying images," Multidimens. Syst. Signal Process., vol. 3, pp. 211--239, 1992.

....approach can be easily extended to cubic trajectories that include the derivative of acceleration. An interesting frequency domain analysis of images containing such high order trajectories is given in [8] It is concluded that, unlike in the constant velocity case where spectral support is planar [21, 10], the spectral support for the case with acceleration and its derivative is truly 3 D. The quadratic motion trajectory model was originally proposed in [11] Later, a similar model was proposed in [7] and extended to trigonometric polynomials. However, both models in [7] were applied uniformly ....

....xyt space. An example of a 3 D trajectory (x(t) y(t) t) of an image point drawn in such a space is shown in Figure 1. Figure 1 To define mathematically a trajectory we use the function c( x; t) such that c is the spatial position at time of an image point which at time t was located at x [10]. c( x; t) describes a 2 D trajectory in the image plane, while (c( x; t) x( y( describes a 3 D trajectory in the xyt space. For each x at time t, the corresponding trajectory starts at time t i (x; t) and ends at time t f (x; t) For 6= t, we can define a subset V( ....

[Article contains additional citation context not shown here]

E. Dubois, "Motion-compensated filtering of time-varying images," Multidimens. Syst. Signal Process., vol. 3, pp. 211--239, 1992.

....may be local, as in the case of pelrecursive methods, or global, as in correspondence methods. Usually, it is assumed in motion estimation analysis that temporal variations between two successive images in a sequence are due to object motion and occlusion effects in the original 3D scene [1] [2]. Under this hypothesis, two similar regions from images at times t and t 1 can be matched using motion models, for example through a displacement and a transformation (e.g. zoom) In practice, the hypothesis of luminance invariance along a motion trajectory rarely holds since there exist other, ....

DUBOIS E. Motion-compensated filtering of time-varying images. Multidimensional Systems and Signal Processing (invited paper), No. 3, pages pp. 211--239, 1992.

.... Gamma1 ; t; t 1g time t, denoted ( d ) t . This can be achieved by explicitly determining d( x; t) at each point x, or implicitly through models such as constant or affine functions on blocks [3] A more general approach is to specify motion trajectories passing through several image frames [4]. Thus, we let c( x; t) be the position at time of an image point located at position x at time t. We can specify the trajectory passing through (x; t) by a parametric function c of a vector p of parameters [1] For example, if we assume constant velocity along the trajectory, c ( x; t) ....

....relate to the entropy of the motion field. On the other hand, for motioncompensated interpolation, the estimated motion field should resemble the true motion as much as possible to maximize the visual quality of the interpolated image sequence. These applications are discussed in more detail in [4] and [1] ACKNOWLEDGEMENT This research was supported in part by a grant from the Canadian Institute for Telecommunications Research under the NCE program of the Government of Canada, and in part by the Natural Sciences and Engineering Research Council of Canada. ....

E. Dubois, "Motion-compensated filtering of timevarying images," Multidim. Syst. Sig. Process., vol. 3, pp. 211--239, 1992.

....to minimize the visibility of flicker at the display. Although interlace is a good technique in the context of analog systems, it usually leads to spatiotemporal aliasing in the presence of vertical motion, which makes subsequent processing, and especially source coding, much more difficult (see [Dubois 1992] for a discussion) For this reason, there has been increasing focus on progressive scanning in recent system proposals. A detailed discussion of sampling structures and video signal sampling is given in [Dubois 1985] Although some standards allow considerable flexibility in the choice of ....

....a tradeoff between coding efficiency and the introduction of artifacts due to noise reduction. Valid assumptions must be made, or the processing may remove desired picture content such as snowstorms, fireworks, etc. Motion compensated processing is required for the most effective noise reduction [Dubois 1992]. However, good image and noise models are necessary. The principle is to filter out those deviations from an assumed image model (e.g. that image intensity is constant along trajectories of motion) that match the characteristics of an assumed noise model (e.g. white noise with a specified ....

Dubois, E. 1992. Motion-compensated filtering of time-varying images, Multidimensional Systems and Signal Processing, 3:211--239.

....that can benefit greatly from the knowledge of motion. In the case of conversion, two scenarios are possible. In one situation a missing image must be recovered. Again, due to the high correlation along motion trajectories, motion compensated interpolation is the most effective tool [53] [15]. In the other situation, only part of an image must be reconstructed. Since some spatial information is available and since motion estimates are occasionally unreliable, methods can be devised which combine smart spatial interpolation [50] 18] with motion compensated interpolation [49] Noise ....

....x; t) mathematically describe a trajectory in the image plane, i.e. let c( x; t) be the spatial position at time of an image point which 1 In this chapter, boldface characters indicate vector quantities, including 2 D spatial coordinates and parameter vectors. at time t was located at x [15]. c( x; t) describes a 2 D trajectory in the image plane, while (c( x; t) x( y( describes a 3 D trajectory in the xyt space. Clearly, there is a unique mapping between the two trajectories. For each x at t, the corresponding trajectory starts at time t i (x; t) and ends at ....

[Article contains additional citation context not shown here]

E. Dubois, "Motion-compensated filtering of time-varying images," Multidim. Syst. Sig. Process., vol. 3, pp. 211--239, 1992.

No context found.

E. Dubois, "Motion-compensated filtering of time-varying images," Multidimensional Syst. Signal Process., vol. 3, pp. 211--239, 1992.

No context found.

E. Dubois. Motion--compensated filtering of time--varying images. Multidimensional Systems and Signal Processing, 3:211--239, 1992.

No context found.

E. Dubois, "Motion-compensated filtering of time-varying images," Multidimensional Systems and Signal Processing, vol. 3, pp. 211--239, May 1992.

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