## Fast Optical Flow Using Cross Correlation and Shortest-Path Techniques (1999)

Venue: | In Proceedings of Digital Image Computing: Techniques and Applications |

Citations: | 5 - 3 self |

### BibTeX

@INPROCEEDINGS{Sun99fastoptical,

author = {Changming Sun},

title = {Fast Optical Flow Using Cross Correlation and Shortest-Path Techniques},

booktitle = {In Proceedings of Digital Image Computing: Techniques and Applications},

year = {1999},

pages = {143--148}

}

### OpenURL

### Abstract

Optical flow or image motion estimation is important in the area of computer vision. This paper presents a fast and reliable optical flow algorithm which produces a dense optical flow map by using fast cross-correlation and shortest-path techniques. Fast correlation is achieved by using the box filtering technique which is invariant to the size of the correlation window. The motion for each scan line of the input image is obtained from the correlation volume by finding the best 3D path using dynamic programming rather than simply choosing the position that gives the maximum cross correlation coefficient. Sub-pixel accuracy is achieved by fitting the local correlation coefficients to a quadratic surface. Typical running time for a 256\Theta256 image is in the order of a few seconds rather than minutes. A variety of synthetic and real images have been tested, and good results have been obtained. 1. Introduction Optical flow or image motion is the displacement of each image pixels in an...

### Citations

1971 | An iterative image registration technique with an application to stereo vision
- Lucas, Kanade
- 1981
(Show Context)
Citation Context ...nstruction. There are several methods of estimating image motion or optical flow [2]. These methods can be divided into correlation-based [1, 14], energy-based [7], phasebased [5], and gradient-based =-=[8, 10, 17, 11]-=- methods. Anandan [1] described a hierarchical computational framework for the determination of dense motion fields from a pair of images. It was based on a Laplacian pyramid and used a coarse-to-fine... |

1737 | Determining optical flow
- Horn, Schunck
- 1980
(Show Context)
Citation Context ...nstruction. There are several methods of estimating image motion or optical flow [2]. These methods can be divided into correlation-based [1, 14], energy-based [7], phasebased [5], and gradient-based =-=[8, 10, 17, 11]-=- methods. Anandan [1] described a hierarchical computational framework for the determination of dense motion fields from a pair of images. It was based on a Laplacian pyramid and used a coarse-to-fine... |

1049 | Performance of optical flow techniques
- Barron, Fleet, et al.
- 1994
(Show Context)
Citation Context ...many applications in image sequence processing, such as robot navigation, object tracking, image coding, structure reconstruction. There are several methods of estimating image motion or optical flow =-=[2]-=-. These methods can be divided into correlation-based [1, 14], energy-based [7], phasebased [5], and gradient-based [8, 10, 17, 11] methods. Anandan [1] described a hierarchical computational framewor... |

451 |
A computational framework and an algorithm for the measurement of visual motion
- Anandan
- 1989
(Show Context)
Citation Context ...obot navigation, object tracking, image coding, structure reconstruction. There are several methods of estimating image motion or optical flow [2]. These methods can be divided into correlation-based =-=[1, 14]-=-, energy-based [7], phasebased [5], and gradient-based [8, 10, 17, 11] methods. Anandan [1] described a hierarchical computational framework for the determination of dense motion fields from a pair of... |

298 |
Computation of Component Image Velocity from Local Phase Information. Int
- Fleet, Jepson
- 1990
(Show Context)
Citation Context ...e coding, structure reconstruction. There are several methods of estimating image motion or optical flow [2]. These methods can be divided into correlation-based [1, 14], energy-based [7], phasebased =-=[5]-=-, and gradient-based [8, 10, 17, 11] methods. Anandan [1] described a hierarchical computational framework for the determination of dense motion fields from a pair of images. It was based on a Laplaci... |

145 |
Optical flow using spatiotemporal filters
- Heeger
- 1988
(Show Context)
Citation Context ...t tracking, image coding, structure reconstruction. There are several methods of estimating image motion or optical flow [2]. These methods can be divided into correlation-based [1, 14], energy-based =-=[7]-=-, phasebased [5], and gradient-based [8, 10, 17, 11] methods. Anandan [1] described a hierarchical computational framework for the determination of dense motion fields from a pair of images. It was ba... |

135 |
On the estimation of optical flow: Relations between different approaches and some new results
- Nagel
- 1987
(Show Context)
Citation Context ...nstruction. There are several methods of estimating image motion or optical flow [2]. These methods can be divided into correlation-based [1, 14], energy-based [7], phasebased [5], and gradient-based =-=[8, 10, 17, 11]-=- methods. Anandan [1] described a hierarchical computational framework for the determination of dense motion fields from a pair of images. It was based on a Laplacian pyramid and used a coarse-to-fine... |

91 |
A computational approach to motion perception
- Uras, Girosi, et al.
- 1988
(Show Context)
Citation Context |

67 |
Optic flow computation: a unified perspective
- Singh
- 1991
(Show Context)
Citation Context ...obot navigation, object tracking, image coding, structure reconstruction. There are several methods of estimating image motion or optical flow [2]. These methods can be divided into correlation-based =-=[1, 14]-=-, energy-based [7], phasebased [5], and gradient-based [8, 10, 17, 11] methods. Anandan [1] described a hierarchical computational framework for the determination of dense motion fields from a pair of... |

45 | Real Time Correlation-Based Stereo: Algorithm, Implementations and Applications
- Faugeras, Hotz, et al.
- 1993
(Show Context)
Citation Context ... bright143 ness and contrast due to the normalization with respect to mean and standard deviation. But direct calculation of ZNCC is computationally expensive compared with SAD or SSD. Faugeras et al =-=[4]-=- used recursion technique and hardware implementation to obtain real time correlation for stereo matching. In [15, 16], Sun presented a method for fast calculations of cross correlation for stereo mat... |

22 | A Fast Stereo Matching Method
- Sun
(Show Context)
Citation Context ...culation of ZNCC is computationally expensive compared with SAD or SSD. Faugeras et al [4] used recursion technique and hardware implementation to obtain real time correlation for stereo matching. In =-=[15, 16]-=-, Sun presented a method for fast calculations of cross correlation for stereo matching purposes. In his case, the searching window was along the 1D epipolar lines. For motion estimation, the search r... |

12 |
Regularised shortest-path extraction
- Buckley, Yang
- 1997
(Show Context)
Citation Context ...tion vectors obtained for neighbouring pixels are more consistent with each other. The best path from left to right through the 3D correlation volume is found by using a dynamic programming technique =-=[3, 6, 9]-=-. The best path gives the maximum sum of the correlation coefficients along the path which satisfies certain connectivity and smoothness constraints. Now we describe an algorithm for the shortest-path... |

8 |
Experiments with symmetrized intensity-based dynamic programming algorithms for reconstructing digital terrain model
- Gimelâ€™farb, Krot, et al.
- 1992
(Show Context)
Citation Context ...tion vectors obtained for neighbouring pixels are more consistent with each other. The best path from left to right through the 3D correlation volume is found by using a dynamic programming technique =-=[3, 6, 9]-=-. The best path gives the maximum sum of the correlation coefficients along the path which satisfies certain connectivity and smoothness constraints. Now we describe an algorithm for the shortest-path... |

6 | Multi-Resolution Rectangular Subregioning Stereo Matching Using Fast Correlation and Dynamic Programming Techniques
- Sun
- 1999
(Show Context)
Citation Context ...culation of ZNCC is computationally expensive compared with SAD or SSD. Faugeras et al [4] used recursion technique and hardware implementation to obtain real time correlation for stereo matching. In =-=[15, 16]-=-, Sun presented a method for fast calculations of cross correlation for stereo matching purposes. In his case, the searching window was along the 1D epipolar lines. For motion estimation, the search r... |

5 |
A dynamic programming algorithm for binocular stereo vision
- Lloyd
- 1985
(Show Context)
Citation Context ...tion vectors obtained for neighbouring pixels are more consistent with each other. The best path from left to right through the 3D correlation volume is found by using a dynamic programming technique =-=[3, 6, 9]-=-. The best path gives the maximum sum of the correlation coefficients along the path which satisfies certain connectivity and smoothness constraints. Now we describe an algorithm for the shortest-path... |

5 | The \Orthogonal Algorithm" for Optical Flow Detection using Dynamic Programming
- Quenot
- 1992
(Show Context)
Citation Context ...pair of images. It was based on a Laplacian pyramid and used a coarse-to-fine matching strategy. Quenot presented an algorithm for the computation of optical flow using orthogonal dynamic programming =-=[12]-=-. The principle was to minimise the sum of square of differences (SSD) between a pair of images. The dynamic programming was performed alternatively on horizontal and vertical image stripes while redu... |