DE BONET J.: Multiresolution sampling procedure for analysis and synthesis of texture images. In SIGGRAPH (1997), pp. 361--368.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....of Electrical and Computer Engineering, University of Minnesota, Minneapolis, MN 55455, USA, guille ece.umn.edu available information from their surroundings. This information can be automatically detected as in [5, 10] or hinted by the user as in more classical texture filling techniques [9, 11, 18]. Several names have been used for this filling in operation: disocclusion in [2, 14] or inpainting in [4, 5, 6] In the context of this paper, and following [5] we shall refer to it as digital inpainting. It turns out that images are not the only kind of data where there is a need for digital ....

J. S. De Bonet, "Multiresolution sampling procedure for analysis and synthesis of texture images," Proceedings of ACM SIGGRAPH, July 1997.

....coordinates. 36 Figure 4.2: Artificial Fur Light Texture (3x3) 36 Figure 4.3: Gaussian Filters in 2D with arrows pointing to derivatives. 39 Figure 4.4: Laplacian Pyramids: original image, pyramids on R, G and B channels. 39 Figure 4. 5: Original tiles are to left, with synthesized textures from [DeBo97] to right. 40 Figure 4.6: Original tiles are to left, with synthesized textures from current reimplementation to right. 41 Figure 4.7: 2D Multi Resolution Synthesis of Fur Image from ACME Facility. 42 Figure 4.8: Silhouette Example 45 Figure 4.9: Tea Pot (left) and silhouette of inner outer ....

....the quadrilinear filtering in the re sampling of the light field itself was sufficient to mitigate discontinuities, however, further research could be applied to this issue. 4. 3 4D Multi Resolution Analysis Synthesis An alternate solution, which has been partially explored, extends the work of [DeBo97] From 2D to 4D. The work of [DeBo97] describes a two phase process for generating new textures from input images. In synthesizing the 2D texture, the input image is first analyzed, identifying the joint occurrence of texture features at multiple resolutions. This is accomplished by constructing ....

[Article contains additional citation context not shown here]

J. S. De Bonet. "Multiresolution Sampling Procedure for Analysis and Synthesis of Texture Images." In Computer Graphics (SIGGRAPH Wesley, Aug 1997.

....means of general triangulation [DmO6] Hop96] A plain colour is assigned to each triangle and progressive refinement is possible by adding more samples. Texture Synthesis and Generation: Simoncelli [Sim98] synthesizes new images by using the jointprobabilities of Wavelet coefficients. De Bonet [DeB97] synthesizes new textures by permuting similar regions based on the joint occurrence of several features through different resolution levels of the Laplacian pyramid. Another texture generation process is based on repeteadly pasting texture pattems over a triangulated mesh [Ney99] Pra00] but, ....

J.S. De Bonet. Multiresolution Sampling Procedure for Analysis and Synthesis of Texture Images. Computer Graphics. Proceedings of SIGGRAPH'97,1997.

....cortex [5] provides a justification for such works on a neurophysiological basis. An alternative approach aiming primarily at texture replication, as opposed to the identification of the perceptual features, has been followed by other authors. Some noteworthy examples are the solutions proposed in [6 12]. Conversely, the field of dynamic texture modeling is relatively under investigated. Dynamic 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 ....

....too simple to be representative of visual processes, provide results comparable in quality to other parametric state of the art techniques suggests some redundancy in the more complex representations at least as far as perceptual features are concerned. The algorithm that inspired the DWT MPTM [6] basically generates a new texture image by shuffling coefficients of the original texture at different spatial resolutions. This shuffling is constrained in two ways: The local characteristics inside a frequency band must be preserved Coefficients in a frequency band depend on the corresponding ....

[Article contains additional citation context not shown here]

J. S. De Bonet, Multiresolution sampling procedure for analysis and synthesis of texture images, in: Computer Graphics, ACM SIGGRAPH, 1997, pp. 361--368.

....receptive fields in primary visual cortex [5] have validated such works. An alternative approach aiming primarily at texture replication, as opposed to the identification of the perceptual features, has been followed by other authors. Some noteworthy examples are the solutions proposed by De Bonet [6], 7] Menegaz [8] Efros and Leung [9] Wei and Levoy [10] Xu et al. [11] and Ashikhmin [12] Conversely, the field of dynamic texture modeling is relatively under investigated. Dynamic textures are usually meant as multi dimensional stochastic processes exhibiting some stationarity over time ....

....of the one performed by the visual system, provide results comparable in quality to other parametric state of the art techniques suggests some redundancy in the more complex representations at least as far as perceptual features are concerned. The algorithm that inspired the DWT MPTM [6] generates a new texture image by shu#ing coe#cients of the original texture at di#erent spatial resolutions. This shuffling is constrained in two ways: The local characteristics inside a frequency band must be preserved . Coe#cients in a frequency band depend on the corresponding ones at ....

[Article contains additional citation context not shown here]

J. S. De Bonet, "Multiresolution sampling procedure for analysis and synthesis of texture image," in Computer Graphics. ACM SIGGRAPH, 1997, pp. 361--368.

....The basic idea behind the algorithms that have been proposed in the literature is to fill in these regions with available information from their surroundings. This information can be automatically detected as in [5, 9] or hinted by the user as in more classical texture filling techniques [8, 13, 27]. The algorithms reported in the literature best perform for pure texture, 9, 13, 27] or pure structure, 2, 3, 5] This means that for ordinary images such as the one in Figure 1, different techniques work better for different parts. In [25] it was shown how to automatically switch between ....

J. S. De Bonet, "Multiresolution sampling procedure for analysis and synthesis of texture im- ages," Proceedings of A CM SIGGRAPH, July 1997.

....segmentation; and (4) shape from texture. Significant progress was made during the 1990s on the first three areas (with shape from texture receiving comparatively less attention) The success in these areas was largely due to learning a fuller statistical representation of filter bank responses [1, 2, 10, 11, 13, 17]. It was fuller in three respects: firstly, the filter response distribution was learnt (as opposed to recording just the low order moments of the distribution) secondly, the joint distribution, or cooccurrence, of filter responses was learnt (as opposed to independent distributions for each ....

J. S. De Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. In Proc. ACMSIGGRAPH, 1997.

....texture over the top of it. These applications all rely on the ability to re synthesize a sample texture to fit a variety of constraints. A number of texture re synthesis methods are described in the literature. One approach is based on searching for specific features in textures [Heege95] [Bonet97] [Porti99] In these methods, the input image is decomposed into a set of features. Statistics about these features are collected, and used to synthesize a new image. One problem with these methods is that they can only recognize a set of features which have been specified in advance. While the ....

J. S. De Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. In Proceedings of SIGGRAPH 1997, pages 361--368, 1997.

....that tends to be impractical for real time applications [Wu00] More efficient techniques tend to properly match texture features ( Por00] measured at different resolution levels ( Bur83] a series of heuristics are used without explicitly derive a real mathematical model. In [Heg95] and [Deb97] impressive results using marginal histograms of image pyramids and maintaining cross scale dependencies are obtained (see also [Bat00a] Bat01b] More recently [Efr99] and [Wey00] pointed out a series of simple but effective techniques showing excellent results on large class of textures. In ....

J.S. De Bonet, Multiresolution Sampling Procedure for Analysis and Synthesis of Texture Images, In Computer Graphics, pp. 361-368, ACM SIGGRAPH, 1997;

....Probably the first step in that direction was taken by Perlin and Goldberg in their Improv system [22] which uses procedural noise to add realistic looking variability to motions. Pullen and Bregler [23] describe a motion synthesis method inspired by De Bonet s texture synthesis algorithm [8]. Their approach decomposes the training data into frequency bands and synthesizes a new sequence, one frequency band at a time. The approach was applied to generate a realistic repetitive motion of a 2D character with three degrees of freedom (a hopping wallaby) Our approach is different since ....

J. S. De Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. In Proc. SIGGRAPH 97, Annual Conference Series, pages 361--368, 1997.

....Texture Synthesis Repetition of Texture Patterns 2.3.1. Texture Synthesis Image synthesis is to produce a new image from an example image, in such a way that: the new image is different enough from the original but seems generated by the same stochastic process that generates the original one [9]. Simoncelli et al. [26] captures random and structured aspects of a texture by means of joint probabilities of Wavelets coefficients. It synthesises a new image from a Gaussian noise image by forcing it to satisfy the given probabilities. De Bonet [9] uses the Laplacian pyramid of the original ....

....process that generates the original one [9] Simoncelli et al. [26] captures random and structured aspects of a texture by means of joint probabilities of Wavelets coefficients. It synthesises a new image from a Gaussian noise image by forcing it to satisfy the given probabilities. De Bonet [9] uses the Laplacian pyramid of the original texture, and computes the joint occurrence of several features through different resolution levels. It considers that there exist interchangeably regions in low frequency levels. The synthesis is done level by level by means of a uniform sampling between ....

[Article contains additional citation context not shown here]

J.S. De Bonet. Multiresolution Sampling Procedure for Analysis and Synthesis of Texture Images.Computer Graphics.Proceedings of SIGGRAPH'97,1997.

....receptive elds in primary visual cortex [5] have validated such works. An alternative approach aiming primarily at texture replication, as opposed to the identi cation of the perceptual features, has been followed by other authors. Some noteworthy examples are the solutions proposed by De Bonet [6], 7] Menegaz [8] Efros and Leung [9] Wei and Levoy [10] Xu et al. [11] and Ashikhmin [12] Conversely, the eld of dynamic texture modeling is relatively under investigated. Dynamic textures are usually meant as multi dimensional stochastic processes exhibiting some stationarity over time ....

....of the one performed by the visual system, provide results comparable in quality to other parametric state of the art techniques suggests some redundancy in the more complex representations at least as far as perceptual features are concerned. The algorithm that inspired the DWT MPTM [6] generates a new texture image by shu ing coecients of the original texture at di erent spatial resolutions. This shuf ing is constrained in two ways: The local characteristics inside a frequency band must be preserved Coecients in a frequency band depend on the corresponding ones at lower ....

[Article contains additional citation context not shown here]

J. S. De Bonet, \Multiresolution sampling procedure for analysis and synthesis of texture image," in Computer Graphics. ACM SIGGRAPH, 1997, pp. 361-368.

....patterns are fused into it to give a more vivid stream. The audio stream can be of arbitrary length according to the need The idea of audio texture is inspired by video textures [7] a new type of visual medium. The latter was proposed as a temporal extension of 2D image texture synthesis [1][2] and is researched in the areas of computer vision and graphics. It is natural to generalize the idea to audio data. Audio data as a signal sequence presents self similarity as a video sequence does. The self similarity of music and audio has been shown in [3] using a visualization method. So ....

J.S. de Bonet. "Multi-resolution sampling procedure for analysis and synthesis of texture images". SIGGRAPH'97. pp. 361-368, 1997.

....textures [26] However recent advances in texture synthesis have produced models that are capable of synthesising natural textures [8] textures that contain both structural and statistical elements. These models are based on the stochastic modelling of various multi resolution filter responses [5], 45] 28] 37] but they do not use third or higher order statistics, and it is undetermined whether the chosen filters are globally optimal for all textures. Julesz [30] suggested there was textural information in the higher order statistics, and Gagalowicz et al. 19] used third order ....

.... model for natural textures [26] However DEC mpp 12000 with 16,384 processors each with 64 kb of memory, which can yield speeds of up to 60 Giga instructions per second new methods based on stochastic modelling of various multi resolution filter responses have produced impressive results [5], 45] 28] 37] Alternatively Popat and Picard [39] successfully used a high order causal nonparametric multiscale MRF model to synthesis structured natural textures. In fact our approach is indicative of theirs, but where they su#ered from phase discontinuity we used our method of local ....

[Article contains additional citation context not shown here]

J. S. De Bonet, "Multiresolution sampling procedure for analysis and synthesis of texture images," in Computer Graphics. ACM SIGGRAPH, 1997, pp. 361--368, http://www.ai.mit.edu/~jsd.

....the interesting ones. The synthesis of realistic textures can be part of the solution. Brick walls, grass, rocks, sand, concrete, vegetation, can be emulated based on a compact model of these textures. Several powerful texture synthesis methods have been proposed over the last couple of years [3, 4, 7, 11, 15, 19]. The realism of synthesized textures has gone up dramatically. With this paper we hope to contribute in a number of respects: # The texture models are very compact, yielding excellent compression. In contrast to several recent methods, the model doesn t contain an example image of the texture. ....

....Nice features of the proposed approach are that the original texture is not needed for synthesis and that no disturbing repetitions of patterns occur, even if large areas of synthetic texture are produced. These aspects may be an advantage with respect to the texture synthesis method of De Bonet [3]. The latter also has difficulties when the raster size of the synthesized image is not a multiple of the period of the textural pattern (Fig. 15) or if the main structural elements of the texture are slightly rotated with respect to horizontal and vertical directions, which have a special status ....

[Article contains additional citation context not shown here]

J.S. De Bonet. Multiresolution Sampling Procedure for Analysis and Synthesis of Texture Images. Proc. Computer Graphics, ACM SIGGRAPH'97, 1997, pp. 361-368.

No context found.

DE BONET J.: Multiresolution sampling procedure for analysis and synthesis of texture images. In SIGGRAPH (1997), pp. 361--368.

No context found.

Jeremy S. De Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. In Proceedings of the ACM Computer Graphics (SIGGRAPH), pages 361--368, 1997.

No context found.

J. S. De Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. In SIGGRAPH '97, pages 361--368, 1997.

No context found.

J. S. De Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. Computer Graphics, 31(Annual Conference Series):361-- 368, 1997.

No context found.

J. S. de Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. In Turner Whitted, editor, SIGGRAPH 97 Conference Proceedings, Annual Conference Series, pages 361--368. ACM SIGGRAPH, Addison Wesley, August 1997. ISBN 0-89791-896-7.

No context found.

Jeremy S. De Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. In Turner Whitted, editor, SIGGRAPH 97 Conference 1997.

No context found.

J. S. de Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. In T. Whitted, editor, SIGGRAPH 97 Conference Proceedings, Annual Conference Series, pages 361--368. ACM SIGGRAPH, Addison Wesley, Aug. 1997. ISBN 0-89791-896-7.

No context found.

J. S. De Bonet, "Multiresolution sampling procedure for analysis and synthesis of texture images," Proceedings of ACM SIGGRAPH, July 1997.

No context found.

DE BONET J. S.: Multiresolution sampling procedure for analysis and synthesis of texture images. In Proceedings of the 24th annual conference (1997), ACM Press/Addison-Wesley Publishing Co., pp. 361--368. 80

No context found.

DE BONET, J. S. 1997. Multiresolution sampling procedure for analysis and synthesis of texture images. In Proceedings of SIGGRAPH 97, Computer Graphics Proceedings, Annual Conference Series, 361--368. 171

No context found.

J. De Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. In Proceedings of SIGGRAPH, pages 361--368, August 1997.

No context found.

Jeremy S. De Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. Proceedings of SIGGRAPH 97, pages 361--368, August 1997.

No context found.

De Bonet J.S., "Multiresolution Sampling Procedure for Analysis and Synthesis of Texture Images", in Computer Graphics, pp.361-368, ACM SIGGRAPH, 1997;

No context found.

J.S. DeBonet. "Multiresolution sampling procedure for analysis and synthesis of texture images." In Proc. of SIGGRAPH97, pp. 361--368, 1997.

No context found.

J.S. De Bonet, "Multiresolution sampling procedure for analysis and synthesis of texture images," Proceedings SIGGRAPH '97, Los Angeles CA, pp 361-368, 1997.

No context found.

J. S. De Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. In ACM SIGGRAPH, pages 361--368, 1997.

No context found.

J. S. De Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. In SIGGRAPH '97, pages 361--368, 1997.

No context found.

J. S. de Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. In Turner Whitted, editor, SIGGRAPH 97 Conference Proceedings, Annual Conference Series, pages 361--368. ACM SIGGRAPH, Addison Wesley, August 1997. ISBN 0-89791-896-7.

No context found.

J. S. De Bonet, "Multiresolution sampling procedure for analysis and synthesis of texture images," Computer Graphics (SIGGRAPH) , pp. 361--368, 1997.

No context found.

J.S. De Bonet. Multiresolutionsampling procedure for analysis and synthesis of texture images. In SIGGRAPH '97, pages 361--368, 1997.

No context found.

J.S. De Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. In SIGGRAPH 97 Conference Proceedings, Annual Conference Series, pages 361--368, August 1997.

No context found.

J. S. de Bonet, "Multi-resolution sampling procedure for analysis and synthesis of texture images," in Proc. SIGGRAPH'97., 1997, pp. 361--368.

No context found.

J.S. De Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. In SIGGRAPH '97, pages 361--368, 1997.

No context found.

J. De Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. In Proceedings of SIGGRAPH, pages 361--368, August 1997.

No context found.

J. de Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. SIGGRAPH, pp. 361-368, 1997.

No context found.

J. Debonet, "Multiresolution sampling procedure for analysis and synthesis of texture images," in ACM SIGGRAPH, 1997, pp. 361--368.

No context found.

J. S. De Bonet, "Multiresolution sampling procedure for analysis and synthesis of texture images," in Proc. ACM SIGGRAPH, July 1997.

No context found.

J. S. De Bonet, "Multiresolution sampling procedure for analysis and synthesis of texture images," Proceedings of ACM SIGGRAPH, July 1997.

No context found.

J. S. De Bonet, "Multiresolution sampling procedure for analysis and synthesis of texture images," Proceedings of ACM SIGGRAPH, July 1997.

No context found.

DE BONET, J. S. Multiresolution sampling procedure for analysis and synthesis of texture images. In Proceedings of SIGGRAPH 97, pp. 361--368.

No context found.

J. S. De Bonet, "Multiresolution sampling procedure for analysis and synthesis of texture images," Proceedings of ACM SIGGRAPH, July 1997.

No context found.

J. S. De Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. Computer Graphics (SIGGRAPH'97), pages 361--368, August 1997.

No context found.

DE BONET, J. S. Multiresolution sampling procedure for analysis and synthesis of texture images. In SIGGRAPH 97 Proc. (Aug. 1997), ACM SIGGRAPH, pp. 361--

No context found.

J. S. De Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. In Computer Graphics, pages 361--368. ACM SIGGRAPH, 1997.

No context found.

Jeremy S. De Bonet. Multiresolution sampling procedure for analysis and synthesis of texture images. Proceedings of SIGGRAPH 97, pages 361--368, August 1997. ISBN 0-89791-896-7. Held in Los Angeles, California.

First 50 documents  Next 50

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