## Proceedings: CVPR |

### From inside the book

Results 1-3 of 76

Page 41

The outcome of the process can be used for both discontinuity detection and

segmentation into shape homogeneous regions. The process is applied to

synthetic noise-free and noisy step, roof, and valley edges, as well as to real

range images. 1 Introduction Shape analysis has been argued to be strongly

dependent on a scale parameter (i.e., scale space [1,5,6,8]). We can identify two

complementary

see Figure l)(see [4] ...

The outcome of the process can be used for both discontinuity detection and

segmentation into shape homogeneous regions. The process is applied to

synthetic noise-free and noisy step, roof, and valley edges, as well as to real

range images. 1 Introduction Shape analysis has been argued to be strongly

dependent on a scale parameter (i.e., scale space [1,5,6,8]). We can identify two

complementary

**approaches**to shape analysis of a non-self-intersecting curve (see Figure l)(see [4] ...

Page 532

3.2 A variance descent

approximation procedure described in the previous sub-section requires an initial

estimate of the different parameters, we are looking for. This sub-section

proposes a method to get such parameters while reducing the search iterations :

Given a window centered in an initial estimate of the corner position given by any

corner measure, or at any arbitrary point chosen interactively, we first consider

the pixels located in ...

3.2 A variance descent

**approach**to initialize the approximation Theapproximation procedure described in the previous sub-section requires an initial

estimate of the different parameters, we are looking for. This sub-section

proposes a method to get such parameters while reducing the search iterations :

Given a window centered in an initial estimate of the corner position given by any

corner measure, or at any arbitrary point chosen interactively, we first consider

the pixels located in ...

Page 556

Abstract We describe an analytical method for recovering 8- D motion and

structure of four or more points from one motion of a stereo rig. The extrinsic

parameters are unknown. The motion of the stereo rig is also unknown. Because

of the exploitation of information redundancy, the

traditional Motion and structure from motion"

less motions are required, and thus more robust estimation of motion and

structure can be obtained.

Abstract We describe an analytical method for recovering 8- D motion and

structure of four or more points from one motion of a stereo rig. The extrinsic

parameters are unknown. The motion of the stereo rig is also unknown. Because

of the exploitation of information redundancy, the

**approach**gains over thetraditional Motion and structure from motion"

**approach**in that less features andless motions are required, and thus more robust estimation of motion and

structure can be obtained.

### What people are saying - Write a review

We haven't found any reviews in the usual places.

### Contents

Detecting Activities | 2 |

ModelBased Recognition | 12 |

Reconstruction of HOT Curves from Image Sequences | 20 |

Copyright | |

76 other sections not shown

### Other editions - View all

### Common terms and phrases

affine affine transformation algorithm alignment analysis angle applied approach axis bitangent calibration camera color component Computer Vision constraint contour coordinate system corresponding curvature curve defined denote derived described detection determine diffusion direction discontinuities disparity distance edge epipolar equation error essential matrix estimation filter frame function Gaussian geometric geometric hashing given global hash IEEE IEEE Trans image plane image points image sequence implementation iterations Kalman filter label linear matching matrix mean curvature measure method minimization motion noise normal object obtained occlusion optical flow orientation pair parallel parameters part-lines particles Pattern pixel position problem Proc projection recognition reconstruction region representation robot rotation scale scale space scene segmentation sensor shape shown in Figure shows smooth solution space stereo stereopsis structure superquadrics surface surface normal surface reconstruction tangent technique tion transformation translation values vector velocity vergence visual hull zero