## Euclidean Structure from Uncalibrated Images Using Fuzzy Domain Knowledge: Application to Facial Images Synthesis (1998)

Venue: | In Proc. International Conference on Computer Vision |

Citations: | 5 - 0 self |

### BibTeX

@INPROCEEDINGS{Zhang98euclideanstructure,

author = {Zhengyou Zhang and Katsunori Isono and Shigeru Akamatsu},

title = {Euclidean Structure from Uncalibrated Images Using Fuzzy Domain Knowledge: Application to Facial Images Synthesis},

booktitle = {In Proc. International Conference on Computer Vision},

year = {1998},

pages = {784--789}

}

### Years of Citing Articles

### OpenURL

### Abstract

Use of uncalibrated images has found many applications such as image synthesis. However, it is not easy to specify the desired position of the new image in projective or affine space. This paper proposes to recover Euclidean structure from uncalibrated images using domain knowledge such as distances and angles. The knowledge we have is usually about an object category, but not very precise for the particular object being considered. The variation (fuzziness) is modeled as a Gaussian variable. Six types of common knowledge are formulated. Once we have a Euclidean description, the task to specify the desired position in Euclidean space becomes trivial. The proposed technique is then applied to synthesis of new facial images. A number of difficulties existing in image synthesis are identified and solved. For example, we propose to use edge points to deal with occlusion.

### Citations

516 |
What can be seen in three dimension with an uncalibrated stering. The second European conference on computer vision italy
- Faugeras
- 1992
(Show Context)
Citation Context ...e to use edge points to deal with occlusion. Keywords: 3D reconstruction, uncalibrated images, image synthesis, representation, fuzzy domain knowledge. 1 Introduction The study of uncalibrated images =-=[2, 3]-=- has many important applications such as the reconstruction of the environment from a sequence of video images where the parameters of the video lens is submitted to continuous modification such that ... |

492 | A robust technique for matching two uncalibrated images through the recovery of the unknown epipolar geometry
- Zhang, Deriche, et al.
- 1995
(Show Context)
Citation Context ...from real images, which are briefly described below. 4.1.1 Establishing point correspondences between images Since we assume affine camera model, we combine the robust matching technique described in =-=[11]-=- and the technique described in [8] to establish point correspondences and determine the affine fundamental matrix between two images. It works reasonably well for facial images differed by a rotation... |

296 | Estimation of relative camera positions for uncalibrated cameras
- Hartley
- 1992
(Show Context)
Citation Context ...e to use edge points to deal with occlusion. Keywords: 3D reconstruction, uncalibrated images, image synthesis, representation, fuzzy domain knowledge. 1 Introduction The study of uncalibrated images =-=[2, 3]-=- has many important applications such as the reconstruction of the environment from a sequence of video images where the parameters of the video lens is submitted to continuous modification such that ... |

249 | View Morphing
- Seitz, Dyer
- 1996
(Show Context)
Citation Context ...es from a small number of real images is important for many applications such as video conferencing and virtual reality. Previous work includes that of Mukaigawa et al. [5] and that of Seitz and Dyer =-=[7]-=-. Mukaigawa et al. assume orthographic projection and generate new images through linear combination of the original images. The coefficients of the linear combination are determined by specifying the... |

66 |
3D motion recovery via affine epipolar geometry
- Shapiro, Zisserman, et al.
- 1995
(Show Context)
Citation Context ...new image. Domain knowledge has been extracted through statistical analysis of 3D range data of real female and male faces. 2 Overview Using affine camera model, we can use the technique described in =-=[8]-=- to determine the affine fundamental matrix F A between two images. Once F A is known, affine structure of the scene can be reconstructed, i.e., given a pair of matched image points (m i ; m 0 i ), we... |

65 | Projective structure from uncalibrated images: Structure from motion and recognition
- Shashua
- 1994
(Show Context)
Citation Context ...narity, collinearity, and cross ratios (ratio of ratios of distances), which is sometimes sufficient for artificial systems, such as robots, to perform tasks such as navigation and object recognition =-=[9, 10, 1]-=-. However, for many other applications, such as virtual reality, we need Euclidean structure, but not the projective structure, because a human being is familiar with Euclidean environments. For those... |

43 |
Navigation using affine structure from motion
- Beardsley, Zisserman, et al.
- 1994
(Show Context)
Citation Context ...narity, collinearity, and cross ratios (ratio of ratios of distances), which is sometimes sufficient for artificial systems, such as robots, to perform tasks such as navigation and object recognition =-=[9, 10, 1]-=-. However, for many other applications, such as virtual reality, we need Euclidean structure, but not the projective structure, because a human being is familiar with Euclidean environments. For those... |

34 | Applications of non-metric vision to some visual guided tasks
- Zeller, Faugeras
- 1994
(Show Context)
Citation Context ...narity, collinearity, and cross ratios (ratio of ratios of distances), which is sometimes sufficient for artificial systems, such as robots, to perform tasks such as navigation and object recognition =-=[9, 10, 1]-=-. However, for many other applications, such as virtual reality, we need Euclidean structure, but not the projective structure, because a human being is familiar with Euclidean environments. For those... |

9 |
Understanding positioning from multiple images
- Mohr, Boufama, et al.
- 1995
(Show Context)
Citation Context ...t is difficult to specify the desired trajectory of the camera in projective space. There are already several pieces of work in the recovery of Euclidean structure from uncalibrated images, including =-=[4]-=-. In this paper, we consider Euclidean structure recovery from affine cameras using fuzzy domain knowledge. Use of affine cameras [6] is motivated by the fact that the depth variation in our domain ap... |

3 | Y.Ohta, “Synthesis of Arbitrarily Oriented Face Views from Two
- Mukaigawa
- 1995
(Show Context)
Citation Context ...is Synthesizing new facial images from a small number of real images is important for many applications such as video conferencing and virtual reality. Previous work includes that of Mukaigawa et al. =-=[5]-=- and that of Seitz and Dyer [7]. Mukaigawa et al. assume orthographic projection and generate new images through linear combination of the original images. The coefficients of the linear combination a... |