## Asymptotically Admissible Texture Synthesis (2001)

Venue: | In International Workshop on Statistical and Computational Theories of Vision |

Citations: | 14 - 1 self |

### BibTeX

@INPROCEEDINGS{Xu01asymptoticallyadmissible,

author = {Yingqing Xu and Song-Chun Zhu and Baining Guo and Heung-yeung Shum},

title = {Asymptotically Admissible Texture Synthesis},

booktitle = {In International Workshop on Statistical and Computational Theories of Vision},

year = {2001}

}

### Years of Citing Articles

### OpenURL

### Abstract

Recently there is a resurgent interest in example based texture analysis and synthesis in both computer vision and computer graphics. While study in computer vision is concerned with learning accurate texture models, research in graphics is aimed at effective algorithms for texture synthesis without necessarily obtaining explicit texture model. This paper makes three contributions to this recent excitement. First, we introduce a theoretical framework for designing and analyzing texture sampling algorithms. This framework, built upon the mathematical definition of textures, measures a texture sampling algorithm using admissibility, effectiveness, and sampling speed. Second, we compare and analyze texture sampling algorithms based on admissibility and effectiveness. In particular, we propose different design criteria for texture analysis algorithms in computer vision and texture synthesis algorithms in computer graphics. Finally, we develop a novel texture synthesis algorithm which samples from a subset of the Julesz ensemble by pasting texture patches from the sample texture. A key feature of our algorithm is that it can synthesize high-quality textures extremely fast. On a mid-level PC we can synthesize a 512 # 512 texture from a 64 # 64 sample in just 0.03 second. This algorithm has been tested through extensive experiments and we report sample results from our experiments. 1 1

### Citations

817 | Texture synthesis by non-parametric sampling
- Efros, Leung
- 1999
(Show Context)
Citation Context ... several algorithms have also been proposed to synthesize textures by matching only local distributions. For example, a non-parametric estimation of MRF models is introduced by (Efros and Leung, 1999)=-=[3]-=-, which can be further accelerated using a tree-structured VQ method (Wei and Levoy, 2000)[11]. 2sThe idea of estimating a non-parametric density can be traced back to Papat and Picard, 1993)[8], wher... |

541 | Image Quilting for Texture Synthesis and Transfer
- Efros, Freeman
- 2001
(Show Context)
Citation Context ...chieve seamless texture synthesis. The method, called patch-based sampling, is described in detail in our technical report [7]. Similar idea has also been independently developed by Erfos and Freeman =-=[4]. Figure 10 shows some of the examples by our patch-based sampling approach 5 . 5 More e-=-xamples are available at the following web site ÛÛÛ� �×�Ó��Ós×Ø�Ø����Ù�ÓÚ�Ð�Ì �ÜØÙÖ��ÅËÊ Ì�ÜØÙÖ��ÀÓÑ�Ô������Ø��Ô... |

450 | Fast texture synthesis using tree-structured vector quantization
- Wei, Levoy
- 2000
(Show Context)
Citation Context ...tributions. For example, a non-parametric estimation of MRF models is introduced by (Efros and Leung, 1999)[3], which can be further accelerated using a tree-structured VQ method (Wei and Levoy, 2000)=-=[11]. 2s-=-The idea of estimating a non-parametric density can be traced back to Papat and Picard, 1993)[8], where they estimate an auto-regression model for ”growing” texture by sampling from a cluster-base... |

400 | Pyramid-based texture analysis/synthesis
- Heeger, Bergen
- 1995
(Show Context)
Citation Context ...statistics. Motivated by the psychology studies, Heeger and Bergen in 1995 proposed a pyramid based algorithm for texture synthesis that can approximately match marginal histograms of filter responses=-=[5]-=-. A mathematical model called FRAME is proposed in (Zhu, Wu and Mumford, 1997) [15]. The FRAME model integrates the filters and histograms into Markov random field models and adopts an accurate but ex... |

308 | A parametric texture model based on joint statistics of complex wavelet coefficients. Int
- Portilla, Simoncelli
- 2000
(Show Context)
Citation Context ...tures by matching joint histogram of a long vector of filter response [2]. (Portilla and Simoncelli, 2000) studied an iterative projection method for matching the correlations of some filter responses=-=[9]-=-. These methods, among many other work in the literature, represent two distinct paths of texture synthesis. The first path learns analytical models of texture using Markov random fields, and synthesi... |

294 |
Markov random field texture models
- Cross, Jain
- 1983
(Show Context)
Citation Context ...terature, represent two distinct paths of texture synthesis. The first path learns analytical models of texture using Markov random fields, and synthesize texture by stochastic sampling from the model=-=[1, 15]-=-. The second path synthesize texture by matching statistics without deriving analytical texture model[2, 9, 14]. The two paths are unified by the equivalence between the Julesz ensemble and FRAME mode... |

243 | Multiresolution sampling procedure for analysis and synthesis of texture image
- Bonet
- 1997
(Show Context)
Citation Context ...ting algorithms explore texture synthesis with joint statistics of filter responses. For example, (De Bonet, 1997) synthesized textures by matching joint histogram of a long vector of filter response =-=[2]-=-. (Portilla and Simoncelli, 2000) studied an iterative projection method for matching the correlations of some filter responses[9]. These methods, among many other work in the literature, represent tw... |

201 | Minimax entropy principle and its application to texture modeling
- Zhu, Wu, et al.
- 1997
(Show Context)
Citation Context ...d a pyramid based algorithm for texture synthesis that can approximately match marginal histograms of filter responses[5]. A mathematical model called FRAME is proposed in (Zhu, Wu and Mumford, 1997) =-=[15]-=-. The FRAME model integrates the filters and histograms into Markov random field models and adopts an accurate but expensive Markov chain Monte Carlo (MCMC) method for texture synthesis. Furthermore, ... |

157 | Lapped textures
- Praun, Finkelstein, et al.
- 2000
(Show Context)
Citation Context ...st with some compromise in effectiveness. This algorithm, first appeared in [13] as a technical report, has been successfully used in texture mapping 3D surfaces (Praun, Frinkelstein, and Hoppe, 2000)=-=[10]. Suppose w-=-e partition the lattice of synthesis £ into a set of Ò disjoint patches, 12sFigure 5: Texture synthesis results. The input samples are of size �� ¢ ��. The synthesized textures are of size ... |

140 |
Visual pattern discrimination
- Julesz
- 1962
(Show Context)
Citation Context ...s can be regarded as the ”same class” of texture for human perception. This theme, studied by psycho-physicists who are interested in the early stage of visual perception, dates back to (Julesz, 1=-=962)[6]. In-=- late 1980’s and early 1990’s, the psychological study also pointed to a conjecture that human texture perception is governed by the empirical histograms of Gabor filtered images, inspired by the ... |

129 | Real-time texture synthesis by patch-based sampling. ACMTransactions on Graphics (TOG
- Liang, Liu, et al.
- 2001
(Show Context)
Citation Context .... We have extended our patch-pasting method to match boundary statistics to achieve seamless texture synthesis. The method, called patch-based sampling, is described in detail in our technical report =-=[7]-=-. Similar idea has also been independently developed by Erfos and Freeman [4]. Figure 10 shows some of the examples by our patch-based sampling approach 5 . 5 More examples are available at the follow... |

72 | Novel cluster-based probability model for texture synthesis, classification, and compression
- Popat, Picard
- 1993
(Show Context)
Citation Context ... 1999)[3], which can be further accelerated using a tree-structured VQ method (Wei and Levoy, 2000)[11]. 2sThe idea of estimating a non-parametric density can be traced back to Papat and Picard, 1993)=-=[8], wh-=-ere they estimate an auto-regression model for ”growing” texture by sampling from a cluster-based model. These methods are good for synthesis (e.g., easy to be implemented and good-quality synthes... |

68 |
Chaos Mosaic: Fast and Memory Efficient Texture Synthesis. Microsoft research paper MSR-TR-2000-32
- Guo, Shum, et al.
(Show Context)
Citation Context ...thesis by Patch Pasting We will now present our algorithm which can be used to synthesize high-quality textures extremely fast with some compromise in effectiveness. This algorithm, first appeared in =-=[13] a-=-s a technical report, has been successfully used in texture mapping 3D surfaces (Praun, Frinkelstein, and Hoppe, 2000)[10]. Suppose we partition the lattice of synthesis £ into a set of Ò disjoint p... |

34 | Exploring Texture Ensembles by Efficient Markov Chain Monte Carlo
- Zhu, Liu, et al.
- 1999
(Show Context)
Citation Context ...ure using Markov random fields, and synthesize texture by stochastic sampling from the model[1, 15]. The second path synthesize texture by matching statistics without deriving analytical texture model=-=[2, 9, 14]-=-. The two paths are unified by the equivalence between the Julesz ensemble and FRAME models [12]. More recently, several algorithms have also been proposed to synthesize textures by matching only loca... |

19 | Equivalence of julesz ensembles and frame models
- Wu, Zhu, et al.
- 2000
(Show Context)
Citation Context ...he second path synthesize texture by matching statistics without deriving analytical texture model[2, 9, 14]. The two paths are unified by the equivalence between the Julesz ensemble and FRAME models =-=[12]-=-. More recently, several algorithms have also been proposed to synthesize textures by matching only local distributions. For example, a non-parametric estimation of MRF models is introduced by (Efros ... |