SBD-Duo: a dual stage shot boundary detection technique robust to motion and illumination effect
- PDF / 2,238,208 Bytes
- 17 Pages / 439.642 x 666.49 pts Page_size
- 77 Downloads / 204 Views
SBD-Duo: a dual stage shot boundary detection technique robust to motion and illumination effect Saptarshi Chakraborty1
· Dalton Meitei Thounaojam1
Received: 24 October 2019 / Revised: 30 July 2020 / Accepted: 20 August 2020 / © Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract In this paper, we propose a novel shot boundary detection technique using gradient and colour information. The gradient similarity and luminance distortion are calculated to measure the contrast and structural changes of each frame including luminance changes. In the proposed system, the effects of the changes in luminance and contrast-structure are integrated via an adaptive method to extract the possible transitions using an adaptive threshold across the videos. In the verification part, CIEDE2000 colour-difference values of the possible transition frames are compared for declaration of abrupt and gradual transitions. Our system takes effectively less computational time to detect abrupt and gradual transition for a video as compared with contemporary solutions. Our proposed system also gives dominate the performance as compared with latest techniques in terms of F1 score using TRECVid 2001 and 2007 selected dataset. We have performed a series of rigorous experimentation to validate our claims. Keywords Gradient · Luminance · Abrupt · Gradual · CIEDE2000 colour-difference · Adaptive threshold · Illumination · Motion
1 Introduction In Today’s era, the Internet is flooded with multimedia data like image and video and it is growing at an exponential rate. It is really a tedious job to organize these huge multimedia data, especially video data, in a systematic and easily accessible fashion. For enabling an efficient browsing of these videos, some techniques or methods should be designed for indexing and retrieving these video data. To overcome this problem, the property and content of the video are utilized for indexing and retrieval which gives rise to the content based video retrieval system. In order to extract the contents of the video, the video is divided into meaningful intra-related frames called shots. In a nutshell, a shot consist of similar contents, either object or features, for a particular consecutive length of frames in a video. Saptarshi Chakraborty
[email protected] 1
Computer Vision Lab, CSE Department, National Institute of Technology Silchar, Silchar, India
Multimedia Tools and Applications
And the process of finding the boundary between two consecutive shots is known as shot boundary detection. Shot boundary detection basically comprises of two types of transitions namely abrupt and gradual transitions. Detection of a sudden abnormal change between two consecutive frames of a video is defined as abrupt or cut transition. Whereas in case of gradual transition, which is basically caused by edit effect in a video, is broadly classified into fade, dissolve and wipes transitions and these types of transitions usually extend to a certain length of frames. Fade transition is an effect o
Data Loading...
