#### DMCA

## MINMAX optimal video summarization (2005)

### Download From

IEEE### Download Links

- [ivpl.ece.northwestern.edu]
- [ivpl.ece.northwestern.edu]
- [ivpl.ece.northwestern.edu]
- [ivpl.eecs.northwestern.edu]
- [ivpl.eecs.northwestern.edu]
- DBLP

### Other Repositories/Bibliography

Venue: | IEEE Trans. Circuits Syst. Video Technol |

Citations: | 23 - 6 self |

### Citations

12155 |
Elements of Information Theory
- Cover, Thomas
- 1991
(Show Context)
Citation Context ...on Optimization Formulations Video summarization can be viewed as a lossy compression process and a rate-distortion framework is well suited for solving this problem. Classical rate-distortion theory =-=[1]-=- characterizes the relation between bit rate and reconstruction distortion for information sources with known distribution (2) (3) (4)sLI et al.: MINMAX OPTIMAL VIDEO SUMMARIZATION 1247 and quantizati... |

1628 |
Error bounds for convolutional codes and an asymptotically optimal decoding algorithm
- Viterbi
- 1967
(Show Context)
Citation Context ...roblem can be solved in stages. For a given current state of the problem, future solutions are independent from past solutions. Exploiting this structure, a dynamic programming (DP) solution based on =-=[30]-=- and [27] is developed. Let the summarization distortion state for the video sequence segment starting with the summary frame selection and ending with the next summary frame be else (9) which is the ... |

1207 | Optimization by Vector Space Methods - Luenberger - 1969 |

227 | Signal compression based on models of human perception - Jayant, Johnson, et al. - 1993 |

218 | Automatic soccer video analysis and summarization
- Ekin, Tekalp, et al.
(Show Context)
Citation Context ...here are also video skim and video retrieval results that incorporate object and semantic level information/preference in generating video skims/summaries. An example is soccer video summarization in =-=[4]-=- and video skim work in [29]. In this paper, we do not cover this type of problems and the formulations do not depend on extracting object level and semantic information. The solutions to the summariz... |

218 | Rate-distortion methods for image and video compression
- Ortega, Ramchandran
- 1998
(Show Context)
Citation Context ..., and a given encoding scheme (for example, motion JPEG, or H.263), the allocation of bits to achieve the desired distortion level can be performed in a rate-distortion optimal manner as described in =-=[23]-=- and [26]. B. Rate-Distortion Optimization Formulations Video summarization can be viewed as a lossy compression process and a rate-distortion framework is well suited for solving this problem. Classi... |

181 |
Bit allocation for dependent quantization with applications to multiresolution and MPEG video coders
- Ramchandran, Ortega, et al.
- 1994
(Show Context)
Citation Context ... no closed form expression of relation between rate and distortion, operational rate-distortion (ORD) theory and schemes are used to achieve good rate-distortion performance. Examples can be found in =-=[25]-=-–[27]. In [17] and [18], we formulate and solve the summarization problem as an operational rate-distortion problem using the average frame distortion as the summarization distortion. However, using t... |

143 | Shot-boundary detection: Unraveled and resolved
- Hanjalic
(Show Context)
Citation Context ...stern.edu). Digital Object Identifier 10.1109/TCSVT.2005.854230 1051-8215/$20.00 © 2005 IEEE from the video sequence, and then selecting “key frames” according to some criterion from each video shot, =-=[9]-=-, [15], [20]. A comprehensive review of past video summarization results can be found in the introduction sections of [8] and [31], and specific examples can be found in [2], [3], [6], [8], [24], [28]... |

132 | Adaptive key frame extraction using unsupervised clustering
- Zhuang, Rui, et al.
- 1998
(Show Context)
Citation Context ..., [20]. A comprehensive review of past video summarization results can be found in the introduction sections of [8] and [31], and specific examples can be found in [2], [3], [6], [8], [24], [28], and =-=[32]-=-. Some of the main ideas and results among the previously published results are briefly discussed next. Zhuang et al. [32] proposed an unsupervised clustering method. A video sequence is segmented int... |

118 | Reliable transition detection in videos: A survey and practitioner’s guide
- Lienhart
- 2001
(Show Context)
Citation Context ... Digital Object Identifier 10.1109/TCSVT.2005.854230 1051-8215/$20.00 © 2005 IEEE from the video sequence, and then selecting “key frames” according to some criterion from each video shot, [9], [15], =-=[20]-=-. A comprehensive review of past video summarization results can be found in the introduction sections of [8] and [31], and specific examples can be found in [2], [3], [6], [8], [24], [28], and [32]. ... |

116 | Multimedia content analysis using both audio and visual clues
- Wang, Liu, et al.
- 2000
(Show Context)
Citation Context ...lecting “key frames” according to some criterion from each video shot, [9], [15], [20]. A comprehensive review of past video summarization results can be found in the introduction sections of [8] and =-=[31]-=-, and specific examples can be found in [2], [3], [6], [8], [24], [28], and [32]. Some of the main ideas and results among the previously published results are briefly discussed next. Zhuang et al. [3... |

107 | Video summarization by curve simplification
- DeMenthon, Kobla, et al.
- 1998
(Show Context)
Citation Context ...rion from each video shot, [9], [15], [20]. A comprehensive review of past video summarization results can be found in the introduction sections of [8] and [31], and specific examples can be found in =-=[2]-=-, [3], [6], [8], [24], [28], and [32]. Some of the main ideas and results among the previously published results are briefly discussed next. Zhuang et al. [32] proposed an unsupervised clustering meth... |

93 |
Rate-distortion based video compression: optimal video frame compression and object boundary encoding
- SCHUSTER, KATSAGGELOS
- 1997
(Show Context)
Citation Context ...iven encoding scheme (for example, motion JPEG, or H.263), the allocation of bits to achieve the desired distortion level can be performed in a rate-distortion optimal manner as described in [23] and =-=[26]-=-. B. Rate-Distortion Optimization Formulations Video summarization can be viewed as a lossy compression process and a rate-distortion framework is well suited for solving this problem. Classical rate-... |

90 |
An integrated scheme for automated video abstraction based on unsupervised cluster-validity analysis
- Hanjalic, Zhang
- 1999
(Show Context)
Citation Context ... then selecting “key frames” according to some criterion from each video shot, [9], [15], [20]. A comprehensive review of past video summarization results can be found in the introduction sections of =-=[8]-=- and [31], and specific examples can be found in [2], [3], [6], [8], [24], [28], and [32]. Some of the main ideas and results among the previously published results are briefly discussed next. Zhuang ... |

80 | A unified rate-distortion analysis framework for transform coding
- He, Mitra
- 2001
(Show Context)
Citation Context ...er of bits needed in encoding a summary frame depends on the coding scheme and desired distortion level for the frame. We assume a bit number is assigned to each frame according to some rate profiler =-=[10]-=- to achieve desired, constant, or near constant peak SNR (PSNR) distortion level among summary frames. Studies show that large fluctuations in PSNR distortion among video frames are very annoying to v... |

56 | MPEG-7 visual motion descriptors - Jeannin, Divakaran - 2001 |

49 | Vibe: a compressed video database structured for active browsing and search
- Taskiran, Chen, et al.
- 2004
(Show Context)
Citation Context ...d video retrieval results that incorporate object and semantic level information/preference in generating video skims/summaries. An example is soccer video summarization in [4] and video skim work in =-=[29]-=-. In this paper, we do not cover this type of problems and the formulations do not depend on extracting object level and semantic information. The solutions to the summarization problem are typically ... |

47 | Video Content Representation Using Optimal Extraction of Frames and Scenes
- Doulamis, Doulamis, et al.
- 1998
(Show Context)
Citation Context ...from each video shot, [9], [15], [20]. A comprehensive review of past video summarization results can be found in the introduction sections of [8] and [31], and specific examples can be found in [2], =-=[3]-=-, [6], [8], [24], [28], and [32]. Some of the main ideas and results among the previously published results are briefly discussed next. Zhuang et al. [32] proposed an unsupervised clustering method. A... |

36 |
Temporal Video Segmentation: A Survey. Signal Processing
- Koprinska, Carrato
- 2001
(Show Context)
Citation Context ....edu). Digital Object Identifier 10.1109/TCSVT.2005.854230 1051-8215/$20.00 © 2005 IEEE from the video sequence, and then selecting “key frames” according to some criterion from each video shot, [9], =-=[15]-=-, [20]. A comprehensive review of past video summarization results can be found in the introduction sections of [8] and [31], and specific examples can be found in [2], [3], [6], [8], [24], [28], and ... |

36 | Katsaggelos, “A review of the minimum maximum criterion for optimal bit allocation among dependent quantizers
- Schuster, Melnikov, et al.
- 1999
(Show Context)
Citation Context ...istortion as The maximum distortion criterion is chosen instead of the average distortion criterion in this case, because it results in more uniformly distributed frame distortion, as is discussed in =-=[27]-=-. The maximum distortion criterion is also found to be a good metric that matches the subjective perception of the distortion. The summarization temporal rate of the summarization process is defined a... |

23 | Constrained Utility Maximization for Generating Visual Skims
- Sundaram, Chang
- 2001
(Show Context)
Citation Context ... [9], [15], [20]. A comprehensive review of past video summarization results can be found in the introduction sections of [8] and [31], and specific examples can be found in [2], [3], [6], [8], [24], =-=[28]-=-, and [32]. Some of the main ideas and results among the previously published results are briefly discussed next. Zhuang et al. [32] proposed an unsupervised clustering method. A video sequence is seg... |

10 | Summarizing video by minimizing visual content redundancies - Gong, Liu - 2001 |

9 |
Time-Constrained Key frame Selection Technique
- Girgenshohn, Boreczky
- 1999
(Show Context)
Citation Context ...each video shot, [9], [15], [20]. A comprehensive review of past video summarization results can be found in the introduction sections of [8] and [31], and specific examples can be found in [2], [3], =-=[6]-=-, [8], [24], [28], and [32]. Some of the main ideas and results among the previously published results are briefly discussed next. Zhuang et al. [32] proposed an unsupervised clustering method. A vide... |

5 | Rate-Distortion Optimal Video Summarization: A Dynamic Programming Solution
- Li, Schuster, et al.
- 2004
(Show Context)
Citation Context ...m expression of relation between rate and distortion, operational rate-distortion (ORD) theory and schemes are used to achieve good rate-distortion performance. Examples can be found in [25]–[27]. In =-=[17]-=- and [18], we formulate and solve the summarization problem as an operational rate-distortion problem using the average frame distortion as the summarization distortion. However, using the maximum fra... |

4 |
Fonctions Aldatories de Seconde Ordre
- Loeve
- 1948
(Show Context)
Citation Context ...imited and the reduced dimensionality makes the covariance matrix estimation from the limited data more accurate. The PCA transform is found by diagonalizing the covariance matrix of the frames [14], =-=[21]-=-, and selecting the desired number of dimensions corresponding to the largest eigenvalues. Therefore, the frame distortion metric is given by (20) where denotes the scaling process, and is the truncat... |

3 |
On linear methods in probability theory
- Karhunen
- 1960
(Show Context)
Citation Context ...ways limited and the reduced dimensionality makes the covariance matrix estimation from the limited data more accurate. The PCA transform is found by diagonalizing the covariance matrix of the frames =-=[14]-=-, [21], and selecting the desired number of dimensions corresponding to the largest eigenvalues. Therefore, the frame distortion metric is given by (20) where denotes the scaling process, and is the t... |

2 |
Schuster “RateDistortion based optimal video summary
- Li, Katsaggelos, et al.
- 2004
(Show Context)
Citation Context ...n background, even though there are a lot of “changes,” the color histogram will stay relatively constant. The computational cost of generating color histogram is also high for larger frame sizes. In =-=[16]-=-, we also experimented with a frame distortion metric that utilizes both color distance and motion activity. For a frame distortion metric that is effective in reflecting the subjective perception of ... |

2 | Rate-distortion optimal video summarization with a bit rate constraint,” presented at the
- Li, Schuster, et al.
- 2004
(Show Context)
Citation Context ...ion of relation between rate and distortion, operational rate-distortion (ORD) theory and schemes are used to achieve good rate-distortion performance. Examples can be found in [25]–[27]. In [17] and =-=[18]-=-, we formulate and solve the summarization problem as an operational rate-distortion problem using the average frame distortion as the summarization distortion. However, using the maximum frame distor... |

1 |
Supervised classification for video shot
- Qi, Hauptmann, et al.
- 2003
(Show Context)
Citation Context ... shot, [9], [15], [20]. A comprehensive review of past video summarization results can be found in the introduction sections of [8] and [31], and specific examples can be found in [2], [3], [6], [8], =-=[24]-=-, [28], and [32]. Some of the main ideas and results among the previously published results are briefly discussed next. Zhuang et al. [32] proposed an unsupervised clustering method. A video sequence ... |