M. Szummer and R. W. Picard. Temporal texture modeling. In IEEE International Conference on Image Processing, 1996.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....The focus of this paper is on editing video based models. 1.2. Other related work Since our emphasis is on time varying textures, we do not address the vast literature on 2D (static) textures here. The problem of modeling dynamic textures has been first addressed by [11] and subsequently by [15]. Procedural techniques have then been proposed by [13] and [18] Time varying texture synthesis algorithms have also been proposed as extensions of 2D texture algorithms (see for waste during visualization and human perception. Since from images alone one cannot disentangle the correct ....

M. Szummer and R. W. Picard. Temporal texture modeling. In Proc. Int. Conf. on Image Processing, volume 3, pages 823--826, 1996.

....a direct global characterization without any prior motion segmentation or without any complete motion estimation in terms of parametric models or optical flow fields. These remarks emphasize the need for the design of new low level approaches in order to supply a direct global motion description, [11, 14, 15]. We follow this point of view and we propose an original method for video indexing with respect to motion content. It uses local non parametric motion related information, and extracts, from temporal cooccurrence statistics, global motion features relative to motion complexity, coherence or ....

M. Szummer and R. Picard. Temporal texture modeling. In Proc. ICIP'96, Lausanne, Sept. 1996.

....textures are usually meant as multi dimensional stochastic processes exhibiting some stationarity over time [13] Some examples are smoke, waves and foliage. This can be regarded as a generalization of the bi dimensional case, where temporal evolution is a feature of the global stochastic process [13 15]. The novelty of our contribution is that we address the problem of modeling a different class of dynamic textures, for which the motion is not an intrinsic property of the considered process, but the result of a continuous change of the point of view. We aim at modeling the motion features as ....

M. Szummer, R. Picard, Temporal texture modeling, in: Proc. of the International Conference on Image Processing (ICIP), Lausanne, Switzerland, 1996.

....some stationarity over time [13] Some examples are smoke, waves and foliage. This can be regarded as a generalization of the bi dimensional case, where temporal evolution is a feature of the global stochastic process. Examples are the solutions proposed by Soatto [13] BarJoseph [14] and Szummer [15]. The novelty of our contribution is that we address the problem of modeling a di#erent class of dynamic textures, for which the motion is not an intrinsic property of the considered process, but the result of a continuous change of the observation point of view. We aim at modeling the motion ....

M. Szummer and R.W. Picard, "Temporal texture modeling," in Proc. of the International Conference on Image Processing (ICIP), Lausanne, Switzerland, 1996.

....some stationarity over time [13] Some examples are smoke, waves and foliage. This can be regarded as a generalization of the bi dimensional case, where temporal evolution is a feature of the global stochastic process. Examples are the solutions proposed by Soatto [13] BarJoseph [14] and Szummer [15]. The novelty of our contribution is that we address the problem of modeling a di erent class of dynamic textures, for which the motion is not an intrinsic property of the considered process, but the result of a continuous change of the observation point of view. We aim at modeling the motion ....

M. Szummer and R.W. Picard, \Temporal texture modeling," in Proc. of the International Conference on Image Processing (ICIP), Lausanne, Switzerland, 1996.

....[13] have been proposed to reduce the computational load. However, the conventional fast methods often converge to a local minimum due to their intrinsic selective nature. The information extracted from the texture analysis of the frame difference signal can be used to alleviate these deficiencies [1, 2, 8, 9, 10]. ME algorithms can be carried out using specialpurpose hardware [17] However, a software solution using general purpose computing platforms is more available. Exploring new ME algorithms is an active area of research, and software solutions have the flexibility to allow experimentation with ....

.... characteristics used in identifying objects or regions of interest in still images, and it can be defined by a set of statistics extracted from the local picture property [20] The texture analysis of the frame difference signals provided to ME algorithms have been proposed in the literature [1, 2, 8, 9, 10]. However, of the various approaches, the method proposed in [1] is chosen because of it reduces the computational load without compromising on motion vectors quality. In an attempt to achieve higher picture quality, the bi directional approach [14] is applied in our implementation of the ME ....

M. Szummer and R. W. Picard, " Temporal texture modeling" , in Int. Conf. on Image Proc., Vol. 3, pp. 823-826, Sept. 1996.

....contents without any explicit prior motion segmentation. Primary work in that direction [NP92] results in the denition of itemporal texturesj which include for instance motions of rivers, foliages, AEames, or crowds. Dioeerent techniques for itemporal texturej feature extraction have been proposed [BF98,FB99,NP92,OHSF98,SP96]. In [OHSF98] descriptors are extracted from the characterization of surfaces derived from spatio temporal trajectories. In [NP92] features issued from spatial cooccurrences of normal AEows are exploited to classify sequences either as simple motions (rotation, translation, divergence) or as ....

Szummer (M.) et Picard (R.W.). Temporal texture modeling. In : Proc. 3rd IEEE Int. Conf. on Image Processing, ICIP'96, pp. 823826. Lausanne, September 1996.

....of images. With a static camera, this stochastic model can be used to extend the sequence arbitrarily in time: driving the model with random noise results in an infinitely varying sequence of images which always looks like the short input sequence. In this way, we can create videotextures [21, 24] which can play forever without repetition. With a moving camera, the image generation process comprises two components a stochastic component generated by the videotexture, and a parametric component due to the camera motion. For example, a camera rotation induces a relationship between ....

.... For the stochastic component, we have access to the wide literature on time series analysis [16, 23] Here, we generally consider autoregressive moving average (ARMA) processes with differencing (ARIMA) The key to this work is the time series model, and in particular the idea of video textures [21, 24]. A video texture is a time series description of a video sequence which 1 Crowd Water Flowers Figure 1: Example image sequences. Crowd : static camera, complex motion. We wish to build a convincing videotexture. See section 2. Water : known camera motion, planar scene. We wish to ....

[Article contains additional citation context not shown here]

M. Szummer. Temporal texture modeling. Master's thesis, MIT Media Lab, Cambridge MA, May 1995.

....pictures, based on color, texture or shape, relatively little of them have focused on using motion to describe video sequences. A recent survey of image and video indexing could be found in [8] An attractive approach consists in using global motion information to video indexing and retrieval [5, 10, 11, 13]. Without needing any prior motion segmentation or complete motion estimation, global motion feature allows to discriminate general types of motion situations. However these techniques remain unsuited to certain sequences when motions are complex, such as human activities. In this paper we ....

M. Szummer and R.W. Picard. Temporal texture modeling. In ICIP'96, September 1996.

....of images. With a static camera, this stochastic model can be used to extend the sequence arbitrarily in time: driving the model with random noise results in an infinitely varying sequence of images which always looks like the short input sequence. In this way, we can create videotextures [21, 24] which can play forever without repetition. With a moving camera, the image generation process comprises two components a stochastic component generated by the videotexture, and a parametric component due to the camera motion. For example, a camera rotation induces a relationship between ....

.... For the stochastic component, we have access to the wide literature on time series analysis [16, 23] Here, we generally consider autoregressive moving average (ARMA) processes with differencing (ARIMA) The key to this work is the time series model, and in particular the idea of video textures [21, 24]. A video texture is a time series description of a video sequence which 1 Crowd Water Flowers Figure 1: Example image sequences. Crowd : static camera, complex motion. We wish to build a convincing videotexture. See section 2. Water : known camera motion, planar scene. We wish to ....

[Article contains additional citation context not shown here]

M. Szummer. Temporal texture modeling. Master's thesis, MIT Media Lab, Cambridge MA, May 1995.

....case, but only a finite length sequence is synthesized after computing the combined MRA tree. Our approach captures the essence of a dynamic texture in some parameters and an infinite length sequence can be generated in real time using the parameters computed off line. Szummer and Picard s work [40, 39] on temporal texture modeling uses a similar approach towards capturing dynamic textures. They use the spatio temporal autoregressive model (STAR) which imposes a neighborhood causality constraint even for the spatial domain. This restricts the textures that can be captured to a large extent. The ....

....a smaller dimensional representation of the image. We incorporate spatial correlation without imposing causal restrictions, as would be clear in the coming sections, and can capture more complex motion. e.g. rotational motion, on which the STAR model is ineffective, taken from the same dataset [40]) 1.2. Contributions of this work This work presents several novel aspects in the field of dynamic (or time varying) textures. On the representation,we present a novel definition of dynamic texture that is general (it captures a wide variety of image dynamics) and precise (it allows making ....

[Article contains additional citation context not shown here]

M. Szummer and R. W. Picard. Temporal texture modeling. In IEEE International Conference on Image Processing, Lausanne, Switzerland, volume 3, Sept 1996.

.... Most existing algorithms model temporal textures by direct simulation; examples include fluid, gas, and fire [23] Direct simulations, however, are often expensive and only suitable for specific kinds of textures; therefore an algorithm that can model general motion textures would be advantageous [24]. Temporal textures consist of 3D spatial temporal volume of motion data. If the motion data is local and stationary both in space and time, the texture can be synthesized by a 3D extension of our algorithm. This extension can be simply done by replacing various 2D entities in the original ....

M. Szummer and R. W. Picard. Temporal texture modeling. In International Conference on Image Processing, volume 3, pages 823--6, Sep 1996.

....described in Section II C is well suited to combine this set of measures and robustly classify image regions into various animal and non animal classes. Note that we are only computing features from still frames and that motion is included explicitly at a higher level. In an alternative approach [27] uses temporal textures for classification, by combining spatial and temporal changes in image sequences. November 13, 1998 DRAFT 11 B.1 Gabor Filter Measures The image (in the spatial domain) is described by its 2 D intensity function. The Fourier Transform of an image represents the same image ....

M. Szummer, "Temporal Texture Modeling," Master Thesis, M.I.T. Media Lab, 1995.

.... cases, all the photos taken by the camera essentially lie in the same orbit of the projective group so that they are related by a simple coordinate transformation [7] Visual synonyms can also occur with patterns, colors, shapes, and textures, including motion patterns or temporal textures [8] [9]. An arrangement of chairs at an outdoor wedding viewed from above may have the same pattern as rows of hedges and flowerbeds in a formal garden. A crowd of people pouring out of a stadium exhibits motion flow similar to candies flowing down a chute in a candy factory. These are examples of events ....

M. Szummer, "Temporal texture modeling," Master's thesis, MIT, Cambridge, MA, May 1995.

....the autoregressive moving average (ARMA) family in Figure 3) 49] was extended for stochastic temporal textures. The standard 2 D model was augmented to form a linear spatio temporal auto regressive (STAR) model, which predicts new image values based on a volume of values lagged in space and time [50]. Using the STAR model, parameters for stochastic temporal textures were estimated, and the motions were resynthesized from the parameters. Resynthesis of motion textures such as steam, river water, and boiling water were found to look natural. These patterns might be thought of as temporal ....

M. Szummer and R. W. Picard, "Temporal texture modeling," in Proceedings ICIP, (Lausanne), 1996. To appear.

No context found.

M. Szummer and R. W. Picard. Temporal texture modeling. In IEEE International Conference on Image Processing, 1996.

No context found.

M. Szummer and R. W. Picard. Temporal texture modeling. In IEEE Int. Conf. On Image Processing, volume 3, pages 823--826, Lausanne, Switzerland, 1996.

No context found.

Martin Szummer. Temporal Texture Modeling. Technical Report 346, MIT, 1995.

No context found.

Martin Szummer. Temporal Texture Modeling. Technical Report 346, MIT, 1995.

No context found.

Martin Szummer and Rosalind W. Picard. Temporal texture modeling. In Proc. IEEE International Conference on Image Processing, volume 3, pages 823--826, 1996.

No context found.

Martin Szummer. Temporal Texture Modeling. Technical Report 346, MIT, 1995.

No context found.

M. Szummer and R.W. Picard. Temporal texture modeling. In Proc. ICIP, 1996.

No context found.

Szummer, M., Picard, R.: Temporal texture modeling. In: IEEE International Conference on Image Processing. Volume 3., Lausanne, Switzerland (1996) 823--826

No context found.

M. Szummer and R. W. Picard. Temporal texture modeling. In Proc. Int. Conf. on Image Processing, volume 3, pages 823--826, 1996.

No context found.

M. Szummer and R. W. Picard. Temporal texture modeling. In IEEE International Conference on Image Processing, 1996.

No context found.

M. Szummer and R. W. Picard. Temporal texture modeling. In IEEE International Conference on Image Processing,vol- ume 3, pages 823--826, 1996.

No context found.

M. Szummer and R.W. Picard. Temporal texture modeling. In Proc. ICIP, 1996.

No context found.

M. Szummer and R.W. Picard, "Temporal Texture Modeling," Proc. Third IEEE Int'l Conf. Image Processing, pp. 823-826, Sept. 1996.

No context found.

M. Szummer and R.W. Picard. Temporal texture modeling. In Proc. of 3rd IEEE Int. Conf. on Image Processing, ICIP'96, pages 823--826, Lausanne, septembre 1996.

No context found.

M. Szummer and R. W. Picard. Temporal texture modeling. In Proc. Int. Conf. on Image Processing, volume 3, pages 823--826, 1996.

No context found.

M. Szummer and R. W. Picard. Temporal texture modeling. In IEEE International Conference on Image Processing, 1996.

No context found.

Martin Szummer and Rosalind W. Picard. Temporal texture modeling. In IEEE International Conference on Image Processing (ICIP 1996.

No context found.

M. Szummer and R. W. Picard, "Temporal Texture Modeling", IEEE ICIP'96, pp.823-826, Sep. 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