Research on Video Abstraction
This paper focuses on designing a system that is capable of abstracting useful video frames for archiving, cataloging, indexing, and editing purpose. Among the different features of video frames, statistics histogram is adopted to detect key frames becaus
- PDF / 1,765,159 Bytes
- 8 Pages / 439.37 x 666.142 pts Page_size
- 104 Downloads / 202 Views
Research on Video Abstraction Linglin Wu, Xiaoyu Wu, Lei Yang and Linwan Liu
Abstract This paper focuses on designing a system that is capable of abstracting useful video frames for archiving, cataloging, indexing, and editing purpose. Among the different features of video frames, statistics histogram is adopted to detect key frames because of its low sensitivity toward motion, low complexity of calculation, and robustness to noise. In addition, cumulative histogram is adopted to detect the edges of video frames due to its lower sensitivity to the motion of objects/camera and illumination variations than statistics histogram. Dynamic threshold-based sliding window is used to detect the shot boundaries and efficiently get the key frames in favor of its representativeness.
Keywords Video abstraction Histogram Shot boundary detection Key frame
83.1 Introduction Along with the rapid development of multimedia and IT technology, multimedia has brought needs for tools that aim at improving contents browsing, searching, and interacting. As the richest style of media, videos are getting more popular than L. Wu (&) X. Wu L. Yang L. Liu College of Information Engineering, Communication University of China, Beijing, China e-mail: [email protected] X. Wu e-mail: [email protected] L. Yang e-mail: [email protected] L. Liu e-mail: [email protected]
Z. Zhong (ed.), Proceedings of the International Conference on Information Engineering and Applications (IEA) 2012, Lecture Notes in Electrical Engineering 218, DOI: 10.1007/978-1-4471-4847-0_83, Ó Springer-Verlag London 2013
675
676
L. Wu et al.
ever; however, how to enable a quick browsing of a large video database has posed challenge to us. In this context, we investigate the abstraction approach, which presents a set of key frames, constituting a salient collection of frames extracted from the video. This set of key frames is remarkably shorter than the original video stream but preserve the representative contents, thus brings efficient content access. Application of video abstraction techniques is wide ranged, such as in data archiving and indexing, film industry, family entertainment, military and public security, medical imaging, image analysis in Aeronautics and Astronautics, and so on. [1]. Storage media such as tape, P2 card store video frames in form of streams, which poses great difficulties to data management and editing. Our goal is to design a system that can efficiently detect the shot boundaries in a certain stream then generates a key frame set for users to locate the video segments they need conveniently.
83.2 Shot Boundary Detection 83.2.1 Shot Boundary In order to generate the key frames set, a segmentation of video content is required before descriptive frames are selected. Thus, shot boundary detection is a crucial technique in video abstraction. Considering massive video, in this paper we present a simple and fast segmentation method based on shot boundary methods. Hence, the change of shot can be determined by detecting shot boundar
Data Loading...
