Shot Boundary Detection with Graph Theory using Keypoint Features and Color Histograms
Loading...
Authors
Lee, Kyoungmin
Kölsch, Mathias
Subjects
Advisors
Date of Issue
2015
Date
2015
Publisher
IEEE
Language
Abstract
The TRECVID report of 2010 [14] evaluated video shot boundary detectors as achieving "excellent performance on [hard] cuts and gradual transitions." Unfortunately, while re-evaluating the state of the art of the shot boundary detection, we found that they need to be improved because the characteristics of consumer-produced videos have changed significantly since the introduction of mobile gadgets, such as smartphones, tablets and outdoor activity purposed cameras, and video editing software has been evolving rapidly. In this paper, we evaluate the best-known approach on a contemporary, publicly accessible corpus, and present a method that achieves better performance, particularly on soft transitions. Our method combines color histograms with key point feature matching to extract comprehensive frame information. Two similarity metrics, one for individual frames and one for sets of frames, are defined based on graph cuts. These metrics are formed into temporal feature vectors on which a SVM is trained to perform the final segmentation. The evaluation on said "modern" corpus of relatively short videos yields a performance of 92% recall (at 89% precision) overall, compared to 69% (91%) of the best-known method.
Type
Article
Description
The article of record as published may be found at http://dx.doi.org/10.1109/WACV.2015.161
Published in: 2015 IEEE Winter Conference on Applications of Computer Vision
Published in: 2015 IEEE Winter Conference on Applications of Computer Vision
Series/Report No
Department
Organization
Naval Postgraduate School (U.S.)
Identifiers
NPS Report Number
Sponsors
Funder
Format
8 p.
Citation
Lee, Kyoungmin, and Mathias Kolsch. "Shot boundary detection with graph theory using keypoint features and color histograms." 2015 IEEE Winter Conference on Applications of Computer Vision. IEEE, 2015.
Distribution Statement
Rights
This publication is a work of the U.S. Government as defined in Title 17, United States Code, Section 101. Copyright protection is not available for this work in the United States.